None of our authors explicitly discusses this criterion, but criticizing it provides Popper with his starting point.
Compare: the verifiability criterion of cognitive significance advocated by the logical positivists.
The positivists were interested in distinguishing empirical claims from "metaphysics." They held that the distinguishing characteristic of metaphysical claims was that there was no way to find evidence to support them. One of A. J. Ayer's examples: "the assertion that the world of sense-experience was altogether unreal" (Language Truth and Logic p. 39). It seems that no sensory experience could establish that all sense-experience is incorrect!
The positivists held that such claims were not cognitively significant (more colloquially, that they were meaningless). If no experience could confirm a claim, then the claim wasn't really an empirical one at all. Although they were not specifically addressing the issue of the bounds of science, they would have been happy to conclude that such a claim could not be part of science.
Popper would probably agree that confirmability is a necessary condition for a theory to be scientific. But he denies that it is sufficient. Popper's suggestion is that confirmation is far too easy to come by, so that it is possible to "confirm" pseudoscientific theories.
Popper's examples: astrology, Marx's theory of history, Freudian psychoanalytic theory, Adlerian psychology. His claim: adherents to these theories see confirmation everywhere. But this is precisely the problem with them: they are too easy to confirm! They are so easy to confirm that in fact they can't be disconfirmed: no matter what happens, there is a way to explain it in terms of the theory.
what does Popper mean by "confirmation," anyway? He seems to have in mind that a theory is "confirmed" if the following two things are true:
T → O (if the theory T is true, then a certain observable consequence O will obtain)
O (O does in fact obtain)
But with theories that "explain everything," it may be that virtually anything we could observe can be "explained" in terms of the theory, that is, we can use the theory (after the fact) to construct a story about why O happened.
One thing this seems to miss is the role of prediction. If the theory is compatible with lots of possible observations, the it doesn't actually predict any of them.
Several possibilities here:
multiple predictions (shotgun approach)
Popper makes the important observation that a test of a theory must be "risky." But why not make that a condition on confirmation, not a rejection of it altogether? That is, perhaps we should deny that T is confirmed if it predicts O and O does occur. That is, perhaps O does not confirm T unless it is true not merely that T → O, but also that if T is not true, then O is highly unlikely (what Elliott Sober calls the "surprise principle").
Popper's picture: a scientific theory can be falsified
T → O
Therefore, not T
But is falsification any more decisive than confirmation? The argument above is deductively valid, unlike the argument form suggested for confirmation above: T -> O, O, therefore T. In practice, though, things are not so straightforward (as Popper was well aware). We don't make predictions on the basis of a theory all by itself, but rather in conjunction with a variety of other assumptions. We'll say more about this later in the semester, but the situation is more like this:
(T and AA and IC) → O
Therefore, not(T and AA and IC)
That is, either not T or not AA or not IC
Where AA stands for "auxilliary assumptions" and IC stands for "initial conditions." When our predictions are not borne out, it could be because the theory is incorrect. But it could also be that our auxiliary assumptions or our beliefs about the initial conditions are incorrect. If every failed experiment refuted a theory, scientific theories would never survive the labs in introductory science courses!
Kuhn: most scientific research is not aimed at confirming or falsifying a theory.
Rather, most research presupposes a theory and tries to use the theory to fill in detailed information that wasn't previously known, or to explain phenomena that it hasn't yet been applied to. In such cases the theory isn't being tested, but rather extended or applied.
Kuhn suggests that the real test of whether a discipline is scientific is whether it has supported such applications and extensions: whether in short it has supported a puzzle-solving tradition.
"A theory or discipline which purports to be scientific is pseudoscientific if and only if:
it has been less progressive than alternative theories over a long period of time, and faces many unsolved problems; but
the community of practitioners makes little attempt to develop the theory towards solutions of the problems, shows no concern for attempts to evaluate the theory in relation to others, and is selective in considering confirmations and disconfirmations."
A discipline is more pseudoscientific the more of the following characteristics it has:
doesn't appeal to laws as universal and unchanging
doesn't attempt to provide explanations and predictions
is not testable (i.e. verifiable or falsifiable)
is not tentative
practitioners do not exhibit "integrity"