by Greg Mayer
Jerry’s post on falsifiability raises interesting questions in the philosophy of science, and I’d like to extend the discussion by referring to a famous incident in the history of science (which is something philosophers of science often do– it’s how they test their ideas). The incident is the discovery of Neptune.
By the end of the 18th century, the orbits of the planets had been fairly well worked out on the basis of observation and Newtonian mechanics. But, by the 1820s, it was evident that Uranus was not following the predicted orbit. The French astronomer Alexis Bouvard perceptively remarked
… I leave it to the future the task of discovering whether the difficulty of reconciling [the data] is connected with the ancient observations, or whether it depends on some foreign and unperceived cause which may have been acting upon the planet.
In the 1840s, two mathematical astronomers took up the challenge, and, hypothesizing that the “foreign and unperceived cause” was an undiscovered planet, attempted to calculate the position of such a hypothetical planet based upon Newton’s laws. They didn’t just speculate about the supposed new planet, though, but also sought to have observatories look for it. The two astronomers were John Couch Adams and Urbain Le Verrier.
On 23 September 1846, at the Royal Observatory in Berlin, Johann Galle, looking where La Verrier had directed him, found the undiscovered planet, which we now know as Neptune. He wrote to Le Verrier two days later:
Monsieur, the planet of which you indicated the position really exists.
Since the 1840s, there has been a dispute as to who deserves the most credit for predicting the existence of the new planet– Adams or Le Verrier. It is clear that it is Le Verrier who actually inspired the successful observations. But the dispute need not detain us– it suffices that the planet was discovered on the basis of someone’s calculations.
So what has this to do with falsification? Well, the problems with the orbit of Uranus meant that there was something wrong with a Newtonian understanding of the solar system. To a naive falsificationist, it would seem that we must conclude that Newton was wrong. But this is not what practicing scientists did. Instead, they sought a way to preserve Newton’s laws, by changing the auxiliary hypotheses used to carry out calculations of predictions under the laws. For the problem was not that Newton’s laws were wrong, but rather that Newton’s laws as applied to the known planets were wrong. The problem therefore could have been not with Newton’s laws, but with our conception of how many planets there were. Schematically
(Newton’s Laws + known planets) predict (orbit U for Uranus); (Uranus does not have orbit U); therefore:
Newton’s Laws, or the known planets, or both, are wrong.
What was falsified was a composite claim– we don’t know which part of the composite is false. (This is an example of what is known as the Duhem-Quine thesis– that hypothese are tested in bundles– mentioned in the comments to Jerry’s piece.)
Why did Adams, Le Verrier, and Galle choose to continue working with Newton’s Laws, rather than abandon them? Because Newtonian mechanics had been a spectacularly successful research program, and it didn’t seem epistemically prudent to throw out the whole thing on the basis of an anomalous orbit, when there were other possible explanations for the anomaly. It seemed reasonable to them (and indeed it was reasonable), to continue with laws that had worked so well in so many circumstances already, rather than concluding they were false.
But isn’t pursuing an escape from falsification a characteristic of pseudoscience? Well, yes it is, if escape from falsification is all your research program consists of. But crucially, Adams and Le Verrier’s escape from falsification was not merely an excuse for the failure of Newton’s Laws to correctly predict the orbit of Uranus. Rather, it was itself an independently testable hypothesis, which, indeed, they (or, more precisely, Galle) tested. This observational test did not depend on the truth or falsity of Newton’s Laws of motion and gravitation, and was thus independent of them. The successful prediction of the existence of Neptune has turned what was potentially a fatal falsification into one of the greatest triumphs of Newtonian mechanics– the discovery of a hitherto unsuspected phenomenon. Newton’s Laws showed what philosophers call “fecundity”.
So, falsifiability is a useful criterion for evaluating scientific claims. But falsifications is a lot harder than the simple logic of ‘A implies B; not B; therefore not A”. There’s not a hard and fast line between scientific and nonscientific claims, but rather a gradation from vigorous, successful research programs, such as evolutionary biology, to degenerate pseudosciences, such as creationism, with things like cryptozoology somewhere in the middle. If all your research program does is move from one escape from refutation to another (it isn’t A, then it’s B; not B, alright, let’s try C; no good?, then how about D; not D,…), then your research program is degenerating. All the signs of vigor (falsifiability, independent testability, fecundity, and others) must be examined in evaluating the epistemic status of a claim or research program.
And a final word, on behalf of Popper. Although he was the falsificationist par excellence, he was not naive, and anticipated many of the criticisms of falsification that later arose. Although he, understandably, emphasized his notion of falsifiability, he began the development of “sophisticated” falsificationism himself.