To be able to tell science from other kinds of study, Karl Popper developed the idea that for a theory to be scientific, there ought to be an experimental way to show that the theory is false based on predictions it makes. He called this criterion “falsifiability” and claimed that it was central to science. To what extent that is true has been a subject of long debate, but the lack of falsifiability is also the basis of attacks on certain well-established scientific theories; the theory of evolution, for example. I think the people who attack evolution in this way make a good point.
When biologists are asked what would cause them to pronounce evolution false, the canonical answer (supposedly first given by JBS Haldane) is a pre-cambrian rabbit. But honestly, even that might not be enough. If we found a fossil rabbit that seemed older than all other life on earth, would that really outweigh everything we know that causes us to believe in evolution? I think not. Instead, we’d spend a lot of time trying to figure out how a pre-cambrian rabbit could have come to be, or, more likely, what kind of mistakes we could have made to think that something is a pre-cambrian rabbit when it actually wasn’t.
What we wouldn’t do is throw out the entire edifice of evolution. Because by this point, it’s more than a theory: it’s a framework for thinking about things and for arranging almost all of our knowledge about biology. You might as well ask what sort of information would invalidate the theory of chemical elements existing. It’s not imaginable, because, again, we are not dealing with individual facts or theories but with ways of thinking and arranging information. That does not make that way of thinking unscientific – it’s just that the criterion for testing individual predictions and huge frameworks might be different.
Some would go even further. Sean Carroll wrote that the idea of falsifiability ought to be retired entirely. He wrote with reference to string theory, which he claims is scientific though not it does not make falsifiable predictions. There was immediate outcry from Carroll’s colleagues, who pointed out just how useful falsifiability has been, and also that the actual criterion philosophers of science talk about when they want to show that science needs a connection with the empirical world is more nuanced.
I agree with much of what Carroll’s critics say. But I do think that a less rigid viewpoint is again useful. Hypotheses that we don’t think are falsifiable or connected to empirical results right now may still be worth putting forward. That is because falsifiability is not a quality that remains the same for ever. For now, I know of no predictions of string theory that everyone agrees are testable. But that may be changing.
I am thinking specifically of the work of people such as Nima Arkani-Hamed to calculate phenomenology questions using string theory tools. Those of us without a string theory background may still not understand what an amplituhedron is, but we can judge it by whether it’s a useful mathematical tool for figuring out how subatomic particles scatter off each other, say at the Large Hadron Collider.
…And it isn’t. At least not yet. But it looks like it might be in the future. And this kind of move from non-testable to testable has a great pedigree because we are not always well equipped to know where science will go. We can easily imagine detectors that are just like what we have, but slightly more sensitive. It is difficult to imagine scientific instruments that do something totally different than what we already have.
For an example of a formerly unfalsifiable idea that is now being tested, we can turn to the world of physical anthropology. Around 40,000 to 50,000 years ago, the amount of cultural artefacts associated with humans increases dramatically. This is usually called the Upper Paleolithic Explosion, and is a huge mystery because it happens about 100,000 years later than the emergence of biologically modern humans does. It’s as if modern humans evolved, then didn’t really do anything new for about 100,000 years, and then suddenly came up with much more complex tools, sewing, burial, art and even music, all at once.
Biologist and anthropologist Richard Klein saw the pattern and came up with an idea: what if it was a single sudden and rapidly spreading genetic mutation that brought about behavioural modernity? There were other explanations offered for the Upper Paleolithic Explosion: the stress of meeting another species of human (Biologically Modern Humans meeting Neanderthals in Europe), a pattern of selective preservation, luck in what turns up in archaeological digs, the emergence of trade, etc. Klein’s theory was completely untestable at the time he put it forward. But now, researchers at the Max Planck Institute for genetics are sequencing archaic homo sapiens genes and can hope to figure out what the difference between them and us is – and whether that difference is something that changed very swiftly 40,000 – 50,000 years ago. If falsifiability came to Richard Klein’s theory, why not to other theories we currently don’t think can be tested? Once again, less absolute rigidity in thinking about scientific theories seems helpful to me.