THE CONVENTIONAL ELEMENT IN SCIENCE

This brings me to the following observation about the tentativity of all scientific activity: almost every scientific hypothesis is backed by promises of ad hoc hypotheses to rescue it from existing refutations.

We all know that scientific hypotheses are tentative in the sense that they can be overthrown by empirical evidence any day now. There are a few writers who do oppose this thesis, among them Pierre Duhem, Ernst Cassirer, Michael Polanyi, Thomas S. Kuhn, and Imre Lakatos. The thesis is usually known as Popper’s, but it really is due to Boyle, Faraday, and Einstein; we may label it as Boyle’s (skeptical) thesis. Now even Duhem et al. agree that facts clash with theory; their point is only that usually theory is rescued ad hoc and so it is naive to believe that what overthrows a theory is its clash with the facts rather than our demand that under certain conditions the clash should overthrow the theory. The question is, under which conditions?

Here I must draw attention to Popper’s specific, and superior, version of Boyle’s rule. When Boyle postulated the rule he considered it as quite unproblematic on the grounds that fact is certain and theory is not, so that if we are certain of the truth of a factual report which contradicts a theory, then, eo ipso, we are certain of the falsity of the theory; ergo, we have already rejected it. Now Popper denies the certitude of factual reports and even enjoins us to try and refute them. So Boyle’s argument will not hold for him. So he needs an added factor, an agreement to reject a refuted theory, and a reason for the agreement: the desire to avoid dogmatism. The convention Popper proposes is to endorse every observation report which has passed tests of agreed severity as true and thereby to have rejected any theory which conflicts with it - until the time when the report is ousted by another report.

Much ink has been spilled over the question, is not Popper a conven­tionalist because of this? Surely in a sense he is, and trivially so. But conventionalists proper are not all those who propose any old convention, but those who propose to stick to theories as conventions and rescue them ad hoc against refuting facts. Here, clearly, Popper is anti-conven- tionalist. But why endorse a convention to defend reports rather than theories? The answer can be found in the previous section: the claim that an ad hoc hypothesis has been turned an observation releases it of its ad hoc nature to a great extent: we prefer ad hocness on the factual rather than the theoretical side: the hie of the ad hoc is the observed fact, whether the one causing trouble and leading to modification or not.

This discussion seems subtle, but I think it can solve an irritating prob­lem, and one which Klappholz and I addressed (Economica, 1959). The problem we were studying was from economic theory. It is well known that the profit-maximization hypothesis was tested in diverse ways, the most obvious of which is the direct questioning of businessmen: do you maximize profits? Now, some economists take this to be a reasonable and proper method, and, as the answer is not universally in the affirmative they consider the theory to be refuted by simple observations. Others may want the theory untestable and untested. Yet the majority, we think, want tests, though not of this type. Why?

Milton Friedman has said that since the profit motive is as abstract as Galileo’s vacuum, to refute it, as to refute Galileo’s vacuum, is besides the point. In other words, we know from the start that the profit motive is not exclusive and not always overriding and we ignore this fact as we ignore friction and the like. In other words, before a test is conducted we may be able to say - more or less (we may change our minds, etc.) - whether its outcome would constitute a refutation or not. And of course, the more we are willing to accept refutations (a priori) the higher the content of our theory, e.g.

So far the view advanced by Klappholz and myself is in line with Popper’s idea of a convention of accepting Boyle’s rule, but it suggests a greater specificity: rather than endorse Boyle’s rule a priori and generally, I can address my opponent in a debate with the following request: do tell me, if I conduct a certain investigation, is there a chance that you will change your opinion? The very meaning of your opinion depends on that answer.

This is no contradiction to Popper’s view, because when you refuse to accept a refutation we may construe your view to mean that it is not contradicted by certain reports. In order to contradict Popper we need examples where clearly a theory does mean to exclude certain reports, these are made, yet the theorizer is quite recalcitrant. Is there such a case? Does such a case conform to the view of Duhem et al/l

Yes, such cases do exist. No, they do not conform to this view. The classical example is Dirac’s case of refusing to admit the truth of reports which conflicted with his theory. The reports, however, were checked and refuted. This is easy to analyze. Dirac’s rejection of the reports was on the mere ground of their conflict with his theory. Everyone would say that this is quite arbitrary and high-handed. Perhaps also dogmatic. But let us look closer: had the reports persisted and Dirac persevered he would thereby perhaps qualify as a dogmatist (perhaps not, depending on what he would have to say for himself). But as the reports swayed his way we can say that his arbitrary and high-handed position was vindicated by the facts. The identification of arbitrariness in some sense with ad hoc will, again, lead us to say that tentative misconduct is not only condoned but at times rewarded. This point, as a methodological point, was first made by Sigmund Freud.

This, however, has nothing to do with the view of Duhem et al. On the very contrary: the view of Duhem et al. presents no problem here, whereas we are cognisant of a problem: was Dirac right? And solve it by saying, yes because he was doing it only for a brief while.

The point which I use contra Popper, and which I am ascribing to Freud, was generalized by Popper himself: whenever we propose a theory to explain the given facts, it usually leads us to think that the facts are slightly different. For example, Newton tells us not of Keplerian ellipses but of perturbed ones, not of Galilean constant gravity but of almost such. Hence, at once the theory is refuted. Yet we give the theory a chance by rechecking the facts.

The question then rises forcefully, how much tolerance should there be? How long should we allow a flagrant and unsettled contradiction between theory and fact?

III.