The Introduction begins with Popper discussing of various issues that have arisen since his publication of The Logic of Scientific Discovery.
The first has to do with the technical terms 'falsifiable' ('empirically refutable') and 'falsifiability' ('empirical refutability'). I first introduced these in Erkenntnis 3,1933, and in Logik der Forschung, 1934, in connection with my solution of the problem of demarcation (discussed at length in Part I, Chapter 2, of the present volume). The problem of demarcation is to find a criterion that permits us to distinguish between statements that belong to the empirical sciences (theories, hypotheses) and other statements, particularly pseudo-scientific, prescientific, and metaphysical statements; but also mathematical and logical statements.
Popper explains the purpose of "falsifiability" as the way to demarcate the difference between the empirical sciences and everything else. Often people misattribute the purpose of "falsifiability" as either a theory of meaning, mistakenly making Popper a positivist, or a theory of truth, or a theory of how knowledge grows.
The problem of demarcation is to be distinguished from the far more important problem of truth: theories which have been shown to be false—as for example the radiation formulae of Rayleigh-Jeans and of Wien, or Bohr's atom model of 1913—can nevertheless retain the character of empirical, scientific hypotheses.
Popper makes explicit two things here. One the problem of demarcation is not the same as the problem of truth. The criteria of falsifiability is not a criterion of truth. And the second thing here is that he considers the problem of truth to be far more important than the problem of demarcation. Often when I encounter people who are not familiar with Popper, they almost exclusively talk about falsification or falsifiability as Popper's main concern or his only contribution. Why is this? Why do people attribute this belief to Popper despite his explicit rejection of it?
Although, following Tarski, I do not believe that a criterion of truth is possible, I have proposed a criterion of demarcation—the criterion of falsifiability. My proposal was that a statement (a theory, a conjecture) has the status of belonging to the empirical sciences if and only if it is falsifiable.
Popper does not think there is a way to have a criterion that differentiates true or false statements. Or true or false theories. The demarcation he proposes is only a demarcation for which statements, theories or conjectures, are scientific. Only if they are falsifiable.
But when is a statement falsifiable? It is of great importance to current discussion to notice that falsifiability in the sense of my demarcation criterion is a purely logical affair. It has to do only with the logical structure of statements and of classes of statements. And it has nothing to do with the question whether or not certain possible experimental results would be accepted as falsifications.
A statement or theory is, according to my criterion, falsifiable if and only if there exists at least one potential falsifier—at least one possible basic statement that conflicts with it logically. It is important not to demand that the basic statement in question be true. The class of basic statements must be characterized in such a way that a basic statement describes a logically possible event of which it is logically possible that it might be observed.
Popper is making it clear that his theory of demarcation is not about finding tests that would be accepted as falsifying a theory. His focus is entirely on the logical structure of a statement. A theory is falsifiable only in the case that at least one basic statement conflicts with it logically.
This comes in conflict with an idea that people often believe refutes Popper. It is usually referred to as the Dunham-Quine thesis. More generally it is a problem of Scientific Underdetermination. But this thesis, despite being true, is not contradictory to Popper, as we can see above this is not a matter of experimental results.
Falsifiability in the sense of the demarcation criterion signifies nothing more than a logical relation between the theory in question and the class of basic statements, or the class of the events described by them: the potential falsifiers. Falsifiability is thus relative to these two classes: if one of these classes is given, then falsifiability is a matter of pure logic—the logical character of the theory in question.
Popper continues,
...falsifiability in the sense of the demarcation criterion does not mean that a falsification can in practice be carried out, or that, if it is carried out, it will be unproblematic. Falsifiability in the sense of the demarcation criterion signifies nothing more than a logical relation between the theory in question and the class of basic statements, or the class of the events described by them: the potential falsifiers. Falsifiability is thus relative to these two classes: if one of these classes is given, then falsifiability is a matter of pure logic—the logical character of the theory in question. That the class of potential falsifiers (or of basic statements) must be given can best be shown by our first example—'All swans are white*.
As I have already said, this statement is falsifiable. Suppose, however, that there is someone who, when a non-white swan is shown to him, takes the position that it cannot be a swan, since it is 'essential' for a swan to be white. Such a position amounts to holding non-white swans as logically impossible structures (and thus also as unobservable). It excludes them from the class of potential falsifiers.
Relative to this altered class of potential falsifiers the statement 'All swans are white* is of course unfalsifiable. In order to avoid such a move, we can demand that anyone who advocates the empirical-scientific character of a theory must be able to specify under what conditions he would be prepared to regard it as falsified; i.e., he should be able to describe at least some potential falsifiers
Popper then turns to a different sense of falsifiability, the type that is not purely logical. How does this work in practice, can a theory ever be so falsified to regard it to be definitively false? Popper says no,
I have always maintained, even in the first edition of Logik der Forschung (1934), and also in my earlier yet only recently published book Die beiden Grundprobleme der Erkenntnistheorie (1979, written 1930-33), that it is never possible to prove conclusively that an empirical scientific theory is false. In this sense, such theories are not falsifiable. 'Every theoretical system can in various ways be protected from an empirical falsification/ (Grundprobleme, p. 353). 'It is always possible to find some way of evading falsification, for example by introducing ad hoc an auxiliary hypothesis . . .' (Logic of Scientific Discovery (L.Sc.D.), p. 42, in the same section in which falsifiability is introduced). 'No conclusive disproof of a theory can ever be produced . . . ' (L.Sc.D., p. 50).
It is clear that Popper never held this second sense of falsifiability, and this is the sense of falsifiability most people believe Popper to hold.
An entire literature rests on the failure to observe this distinction. It is often said that my criterion of demarcation is inapplicable because empirical scientific theories cannot be definitively falsified. Less importantly, it is often said (see section IV below) that the discovery of the unfalsifiability of scientific theories in the second sense is an achievement that contradicts my theory, despite the fact that I myself have pointed this out over and over again.
in Page xxxv Popper writes,
The special thing about human knowledge is that it may be formulated in language, in propositions. This makes it possible for knowledge to become conscious and to be objectively criticizable by arguments and by tests. In this way we arrive at science. Tests are attempted refutations. All knowledge remains fallible, conjectural. There is no justification, including, of course, no final justification of a refutation. Nevertheless we learn by refutations, i.e., by the elimination of errors, by feedback. In this account there is no room at all for 'naive falsification'.