Prove It

To the Editors:

…I cannot let pass what I consider an error in Peter Singer’s review of Popper’s contribution to the problem of induction [NYR, May 2].

Popper showed that induction cannot serve to prove a scientific theory. Nothing can: theories cannot be verified, they can only be falsified. He did not rule out induction as a way of arriving at scientific hypotheses, all he said was that induction cannot be used as a way of proving them. In so doing he relegated induction to the same nonlogical status as intuition.

Peter Singer seems to find it difficult to accept the idea that scientific theories are, and must remain, hypothetical in character. “Unfortunately, we still have to act”—he complains. True; but not valid as an objection to Popper’s argument. It would be valid only if one needed logical proof as the basis for decision making. Popper has shown that this position, which he labelled “comprehensive rationalism,” is untenable for the simple reason that often when decisions are unavoidable logical proof is not available. In the absence of proof, theories which can be falsified but have not been falsified must suffice.

George Soros

New York City

Peter Singer replies:

I do not find it in the least difficult to accept the idea that scientific theories are hypothetical. My point was only that this in no way solves the problem of induction. As Popper himself says in Objective Knowledge, the problem is one that applies to inductive arguments that are held to make a conclusion probable, as much as to inductive arguments designed to make a conclusion certain. Popper, in his attempt to solve the problem, is unable to show why it is to any extent more probable that I will descend safely to street level from my seventeenth-floor apartment by taking the elevator than by jumping from the window. But if I wish to get to the street, I have to act—and that is why we need an explanation of how past experiences justify inferences, probable or certain, about the future. Yet Popper says that the fact that a theory has been corroborated in the past “says nothing whatever about future performance.” Hence Popper has not cleared up the problem of induction.