Skip to main content

Karl Popper liked to say that he had solved the problem of induction, and his solution was startling. He was a Viennese contemporary of the Vienna Circle, never a member, and one of its sharpest critics, though the Circle published his first book in its own series in 1934. His most famous idea, that scientific theories must be open to refutation, is widely known. Less known are the bolder claims around it.

Science without induction

Hume, Popper agreed, had shown that induction cannot be justified. Where others looked for a rescue, Popper denied that science needs induction at all. Science, on his account, proceeds by conjecture and refutation: scientists propose bold guesses, deduce what they imply, and try hard to show them false. The logic of testing is deductive. If a theory implies an observation and the observation fails to appear, the theory is refuted by a valid argument.

What, then, should we say about a theory that survives? Popper called it corroborated. Corroboration is a report of how a theory has fared against severe tests, not a sign that it is probably true. He insisted on the difference. A bold theory says a great deal, and so is improbable from the start; that is exactly why it is valuable, and its survival tells us it has withstood our best attempts to break it, nothing more.

The practical objection

Critics thought this left a gap. The American philosopher Wesley Salmon pressed it in 1981. An engineer choosing between a well corroborated theory of materials and a wild untested one will rely on the first when building a bridge. Why? If corroboration is only a record of past performance, it seems to give no reason to expect future success. Popper's account struggles to explain why we rely on tested theories in practical action, and many critics concluded that some form of induction had slipped back in through the side door.

Rescues: honest and dishonest

Any theory can be saved from a failed test by adjusting something else: blaming the instruments, adding a new assumption, redefining a term. Popper called the abuse of this freedom the conventionalist stratagem: saving a theory by moves that make it immune to refutation. He did not forbid all rescues, because some lead to discoveries. His rule was that an auxiliary hypothesis is acceptable when it makes the theory more testable rather than less, so that the rescued theory sticks its neck out in some new place. A rescue that only explains away the awkward result, and predicts nothing further, he called ad hoc.

Closer to the truth

Popper also wanted to say that science makes progress even though every theory may eventually be refuted. Newton's physics was false, yet it seemed nearer the truth than what came before. He called this nearness verisimilitude, a theory's closeness to the truth, and in 1963 offered a definition based on comparing the true and false consequences of theories. In 1974 the philosophers David Miller and Pavel Tichý independently proved that on Popper's definition no false theory could ever be closer to the truth than another false theory. Since every past theory in physics was false, the definition could not capture the progress it was built to explain. Popper accepted the proof, and the search for a better definition continues.

Chance and natural selection

Two further difficulties concerned what falsifiability could cover. Much of science makes claims about chance: a coin lands heads half the time, a gene appears in a quarter of offspring. No finite run of results strictly contradicts a probabilistic hypothesis, since even a thousand heads in a row is possible. Popper proposed a methodological rule: treat sufficiently improbable results as refutations.

The second difficulty was evolution. In his autobiography of the 1970s Popper described Darwinism as a metaphysical research program rather than a testable theory. In 1978 he published a correction, concluding that natural selection can be tested after all, at least in many of its forms.

Popper drew a wider moral. The critical attitude that keeps science honest, the readiness to say "I may be wrong," was for him also the root of a free society, in which no ruler or doctrine stands above criticism.

Worked example: the shrinking arm

In 1887 Albert Michelson and Edward Morley looked for the motion of the Earth through the ether, the medium that light waves were thought to travel in, and found nothing. Two years later the Irish physicist George FitzGerald, and in 1892 the Dutch physicist Hendrik Lorentz, proposed that objects moving through the ether shrink slightly along their direction of travel, by exactly the amount needed to hide the effect. Popper cited the contraction hypothesis as a textbook ad hoc rescue: it seemed to explain away one awkward result and nothing more. Later philosophers, among them Adolf Grünbaum, objected that it did have testable consequences, and in 1932 an experiment by Roy Kennedy and Edward Thorndike tested a version of the apparatus that contraction alone could not account for. The case cuts both ways. It shows why Popper's worry about convenient rescues matters, and it shows that judging whether a rescue is ad hoc is harder than it looks at the moment it is proposed.

Quiz

5 questions · Practice: scored, not saved

1. When did Popper accept an auxiliary hypothesis that rescues a theory?
2. What did Popper mean by corroboration?
3. Why are probabilistic hypotheses hard to falsify?
4. How did Popper respond to Hume's problem of induction?
5. What was Wesley Salmon's objection to Popper?
6. What did Popper call the conventionalist stratagem?
7. What did Popper conclude about natural selection in 1978?
8. What did Miller and Tichý prove about Popper's definition of verisimilitude?

CivilSimian.com created by AxiomaticPanic, CivilSimian, Kalokagathia