POPPER IS AMBIVALENT REGARDING GOODMAN’S PROBLEM

For anti-instrumentalists like Popper or like myself it is almost possible to relegate Goodman’s question - what distinguishes the rational forecast from the irrational one? - from the realm of theoretical science to the realm of rational technology.

This is not to say that Popper does relegate reliability to the domain of rational technology. All in all, I do not think he distinguishes clearly enough between science and technology, except when he criticizes in­strumentalism. He does say, in technology we may, for technical reasons, prefer to employ a refuted theory instead of an unrefuted theory, so that technology differs from science, where (refutable but) unrefuted theories are always preferenced to refuted ones. This, however, does not make him raise the question, when is the employment of a refuted theory in tech­nology rational, when not. On the contrary, he either ignores the question of reliability altogether or he tackles it in general within science - not within rational technology as distinct from pure science. Clearly Popper is ambiguous here.

In his chapter on the empirical basis of science in his classical Logik der Forschung (i.e. logic of research, which he translated under the title The Logic of Scientific Discovery), Popper speaks of empirical information as the basis of science in general. He rules - quite incorrectly, historical examples show clearly - that whenever we have empirical information conflicting with theoretical hypotheses we must relinquish the hypotheses and endorse the information. Otherwise, he says, we may become dog­matic and there must be a rule, he says - quite dogmatically, I think -, to barr dogmatism.

Popper, then, accepted facts as a basis, but as a merely temporary basis. He likens the foundations of science not to concrete poured on a rock but to piles drilled in a swamp. I am afraid I am not clear about this metaphor. In particular, since Popper’s chapter on the empirical basis discusses mainly refutation, not confirmation, I do not see how this exactly relates to the question of reliability - of theories or of general facts. I suppose that when Popper speaks of the empirical basis of science he does not speak of the reliability of theories. Rather, he speaks there of the growth of science - he says at the end of his chapter “the bold structure of its science’s theories rise, as it were, above a swamp. It is like a building erected on piles”. As long as he speaks of progress, and as long as he keeps clear the idea that progress is fed on refutations, then, it is fairly clear that he is asking about the reason for the acceptability of refuting information. He says, we cannot prove its truth, but we agree to accept it as true as long as it is not refuted, in order to oust dogmatism.

In so far as this is acceptable, and let us consider it acceptable for now, clearly it is a solution to the problem of induction for empirical generali­zations (as scientific information is always generalized: unrepeatable data are not allowed within science) within the field of pure scientific research. The scientist takes a hypothesis to be refuted, he looks for an alternative to it, on the basis of his blanket agreement to accept as reliable all re­futing generalizations. This agreement, says Popper, is made in order to keep science going: violating it opens the door to policies which may bring about stagnation through dogmatism.

This is all rather drastic. There is no need to fear dogmatism so: the dogmatism of Newton, or of Heisenberg, did not kill science. There is no need for a blanket agreement, since we can report an occasional reasonable rejection of attested empirical information for the sole pur­pose of rescuing a theory. Dirac’s case is celebrated; but already Prout faced the same situation when he proposed his hypothesis: many atomic weights were reported then to be not whole numbers; he got rid of some of them by better experiments, others by modifying some subsidiary hypotheses (he replaced the formula for water from HO to H₂O); and he expressed high hopes about the rest. Stas refuted, or so he thought, some of the hope; yet the hope was reasonable.

We must, to note the logic of our situation, observe a few trivial points. Goodman’s question is, what differentiates a rational prediction from an irrational one or, which is the same for our present context, a rational generalization; and this answer must be confined, as yet, to the domain of pure science. To be brief, let us introduce Popper’s technical term “corroborated” to mean “well-tested and as yet unrefuted” (the word has a different sense in legal terminology, and the ordinary sense is either the legal one of that of confirmation or even strengthening; Poincare, we remember, linked confirmation to refutation, the way Popper links cor­roboration to refutation). Popper’s view may be an answer to Goodman’s question, and as follows. It is rational to rely on all and only corroborated generalizations. If, however, Popper were to allow for the rational rejec­tion of even one corroborated generalization then he would thereby allow us to raise afresh Goodman’s question. It seems to me clear that this is indeed Popper’s view, and that he does answer Goodman’s question, all be it erroneously. But suppose that my reading of Popper is false. The question, then, is, how else does Popper answer Goodman’s question? If my reading of him is false, I contend, then, he leaves Goodman’s question unanswered.

So much for Popper on the reliability of empirical generalization. As to scientific theories, Popper says they can be corroborated too: we may fail to refute them. Literally this is doubtlessly trivially true. Does that make reliance on them rational? Popper is unclear here. Supposing it is, is that rationality obligatory? Again, no clear answer. Practically all philosophers occupied with some exegesis on Popper’s text (except my­self) say, yes, Popper does find it rational and hence obligatory to believe a corroborated theory. It seems to me that Popper says, belief in a corrob­orated theory is rational but not obligatory. This, of course, would be trivial if belief in any unrefuted theory is rationally permissible but not obligatory: this is simply permissiveness. Thus, the view that belief in a corroborated theory is permissible may be an austere view if it implies that it is irrational to believe an uncorroborated view, alternatively, it may be the permissiveness and that would make corroboration rather irrelevant to belief.

To conclude I find Popper’s theory of corroboration confusing. It may be read, I suggest, as one or the other theories:

(1) A corroborated theory must be accepted until it be refuted or re­placed by a more testable one. That is, assent must be given to a corrob­orated theory.

(2) A corroborated theory need not be accepted, and any doubtful theory (including a corroborated one) may be accepted. That is, assent is always optional, except in cases of refuted theories.

(3) A corroborated theory need not be accepted, but an uncorrob­orated (and a refuted) one must be rejected. That is to say, we may always suspend judgment but must suspend judgment on all theory except the corroborated theory.

Each of these theories seems to me objectionable, thought, of course, I find the first and most constraining to be the most objectionable. I can find with ease quotations from Popper supporting each of these three readings, but rather than waste time on hermeneutics I should leave it to him to clarify his position if he is interested. If he’s not interested, then, I suggest, hermeneutics is out of place: it is unbecoming to read into a text what an author has left open.

III.