UNITY OF SCIENCE AS A DICTATOR OF UNANIMITY ON ALL QUESTIONS

When we distinguish the rational from the capricious, from the whimsical - even merely from the private and personal - we scarcely do more than accept the minimum program for rationality, the desideratum for a doc­trine of rationality accepted by everyone.

Justificationism, I claim, totally polarizes the subjective from the objec­tive, the non-rational from the rational. It leaves no room for any middle ground of partial rationality of opinion, of partial reasonableness, of degrees of rationality; similarly, it leaves no middle ground for any degree of rationality of a decision, in the realm of morality and responsibility, in the realm of daring and courage.and boldness, in the realm of putting polit­ical and social proposals to critical debate and attempted implementation; there is no room within justificationism for the partial justification even of a scientific project, be it a theoretical exploration or a design of an experiment or a sociological survey or anthropological fieldwork.

This contention of mine is open to a number of criticisms w}iich I shall soon expound and deflate. First let me show how very unreasonable the polarization is, especially in that it makes rationality so uniform that no room is left for diversity. Both the Middle Ages and the 18th century, at opposite poles, suffered from this.

One of the central tenets of Mediaeval philosophy was that man is irrational; for all men sin, and the wages of sin are infinitesimal in comparison with both the punishment sin entails and the rewards of righteousness.

Nowadays the leading theory of rational conduct is that of Bruno de Finetti, the theory of subjective probability and of revealed preferences. The theory is very simple. It grants, first, that there is no universal scale of assessing the probability of a given prediction in the light of given experiences; it assumes, second, that everyone adopts his own scale and everyone acts on it. If any person claims to have no judgment of the likelihood of a prediction, we may in principle show he conceals his opinion by challenging him to a wager. This will call him to act on his assessment, and since he will act rationally he will reveal it. As long as we are not concerned with the cause of the concealment, be it deception or self-deception, the theory seems to cover all possible cases - in princi­ple, that is.

This view of de Finetti is now endorsed by Rudolf Carnap and many other leading thinkers.¹⁹ It represents the pole that we are all rational. Contrary to it, and contrary to the Mediaeval idea, as well as the 18th- century idea, I suggest, as a matter of sheer commonsense, that we are all partially rational, that there are degrees of rationality, that there are cases of better and poorer rationale, that there are degrees of responsi­bility, that we are sometimes more, sometimes less reasonable, responsi­ble, deliberate.

We have already met the opposite idea, explicitly stated by Bacon, that competing schools, doubts, controversies, are symptoms of lack of reason. Bacon also stated in the preface to his Great Installation, you can’t criticize a man and yet respect him. This is in character, yet fantastic. Adam Smith, in his obituary of his friend David Hume says, of his philosophy I shall say nothing since those who agree with it admire it, and those who do not, despise it. I think today most of us respect and yet dissent from Hume. Benjamin Franklin exhibits in his Autobiography a curious ambivalence to debate. He enjoyed debate in his youth, but took pride in not having argued with his scientific opponents who lost ground even without a battle. Much as we disagree with Adam Smith, we usually agree with Franklin. At least we have a problem here.

In a well-known paper²⁰ Stephen Toulmin rejects the thesis that Priestley was foolish to reject Lavoisier’s doctrines and seeks to show that he was justified. The criterion of choice of a hypothesis, says Toulmin, is that of simplicity; and which of the available doctrines was simpler was an open question for quite a while.

This is a nice point. Toulmin agrees with Bacon and his followers about clear cut decisions and the inadmissibility of dissenting from them once they were computed; he only claims that the interim period of computa­tion may be long. But Toulmin is even historically mistaken: it was Kirwan, not Priestley, who held the criterion of simplicity; and applying the criterion he soon gave up his phlogistonist views and joined Lavoisier; Priestley had different criteria, which are too complicated to present here.²¹ Whatever the criteria within science, even when one accepts them within science, and even when their application is unproblematic, dissent may still be sustained, say within metaphysics.

Examples: Faraday agreed that Newtonian mechanics had been as well confirmed as possible, yet he rejected it on metaphysical grounds: he rejected action at a distance on the basis of the idea of conservation of force. Einstein agreed that quantum theory is highly satisfactory as a consistent explanation, yet his determinism made him reject it. Were these people unreasonable? By implication, Toulmin’s paper comes dangerously close to saying, yes. I think it is not too much to suggest, intuitively if you like, that they were both reasonable, though most of us would side with Faraday more easily than with Einstein.

The doctrine of proof requires that all arguments be quickly terminable. The verification principle is but a variant of it. To claim that instead of verification we have only probability looks like partial rationality, but this is not so: we all have to endorse the most probable hypothesis, and to the degree that it has been rendered probable. The question is, of course: What hypothesis, among competing ones, is the most probable? To this there is no generally accepted answer, but since inductive philosophers are still struggling to find such a universal qriterion, claiming to be able to find it in the relatively near future, we should not press this question now; suffice it that they claim that in science unanimity is achieved over any specific case of choice of hypothesis. Of course, the unanimity thus re­ported is a myth too, but let us not press this point either. Rather, we can ask: What does the endorsement of a hypothesis amount to? Endorsing the most probable hypothesis, most inductive philosophers claim, is the bare recognition of the (logical) fact that a given hypothesis is the most probable among the class of competing hypotheses in the light of experi­ence and thus the recognition of its rational preferability.²² Does this solve any problem? Does Einstein’s recognition of the merits of quantum theory amount to endorsement of that theory? In the accepted sense it does; in fact it does not.

But perhaps we can find two equally probable hypotheses? Will not this, at least, justify diversity of opinion?

We barely allow for diversity this way and the choice looks to us dangerously arbitrary. So we add a criterion of choice: of two equiproba­ble hypotheses choose the simplest. And farewell to diversity.

Ambivalence plays its tricks here; diversity looks arbitrary and hence irrational since we have contrasted caprice and idiosyncrasy with objec­tivity and universality and rationality. Once we have done that we cannot ameliorate matters. We have to start afresh: we must identify caprice and

lack of rationality but not conclude from that that rationality is com­pulsive to all rational men. How can this be done is a question which I shall come to in the next paragraph. First I wish to state that - paradoxi­cal as it sounds - we do in common experience identify caprice with lack of rationality yet refuse to view all diversity as caprice; hence we refuse to see rationality and caprice as contrasts though we identify caprice with lack of rationality.

The paradox, in other words, is a logical error; identifying the minimal points of two scales is not identifying the two scales. Even if each is composed of two points each may have only one point in common with the other. It may happen that all good money is genuine, and that all forged money is bad, yet from this it does not follow that all bad money is forgery! It a fortiori does not follow from all irrationality is sheer caprice that all rationality is totally binding. We may introduce leeway, and it is commonsense that we should do so.

Even on a question on which we have utterly no evidence, the classical doctrine of rationality allows no diversity! Some say, have utterly no judgement on matters concerning which you have no proof, and with no judgement engage no action.

The intuitive clash between universality and diversity was felt in differ­ent ways by different thinkers. Let me mention one. William James felt it would be a bigotry to consider his own brand of Christianity the only rational one or to say that choice of religion is totally arbitrary. He developed pragmatism in order to permit everyone to choose a religion, that is to say, to offer more than one religion as rational choices. He ended up by a criterion which, if it were operative at all, would compel each one to choose one and only one religion - the most useful one. 1 suppose it is the religion of his environment, but I am not sure.²³

The same may be said of Carnap's principle of tolerance. Carnap first claimed that metaphysical problems - and solutions to them - are meaning­less: he then claimed that they may be rendered meaningful by translating them to questions concerning the language-systems we may employ. Now, it seems, the question of choice between competing metaphysical systems gets translated or transformed into questions concerning the choice of language-systems. This question Carnap has answered with a principle he called “the principle of tolerance": everyone has the complete right to choose his own language-system. Now this principle may be a principle of indifference, in which case there is little value in the tolerance employed: we do not have to display much tolerance to allow people to choose be­tween alternatives of equal merit. If, on the contrary, one language-system is preferable to another, tolerance (which we all take for granted anyway - otherwise there would be no problem of choice) is no solution to the problem of choice: we need criteria and rules of rational discussion to facilitate rational choice i.e. the choice of the best language-system. We are caught here in the same net. Either there is no rational choice of a framework - the framework is arbitrary whether it be a religion or a language-system (a metaphysics) - or there is a criterion of rational choice and it imposes on us a unanimity; all who share it are rational and all dissenters are irrational.

One final example, an intriguing one, and with a note of depair. R. G. Collingwood, Speculum Mentis or the Map of Knowledge of 1924; chapter on History, section on the World of Fact as Absolute Object. I quote:

In science the attempt is made to bridge the gulf between God and the world. Instead of God we find the concept or law, which because it is the law of the world no longer stands outside the world. Here what was transcendent has become immanent. But in this very success, failure is revealed; for it becomes plain that the immanence is a false immanence [for the following reason: in science the fiction is held that] the world is not a world of individuals but a world of particulars, and because the concept [i.e. the law] is indifferent to the various particulars in which it is embodied, their diversity remains meaningless and the world is to that extent a chaos... the failure of religion is repeated and the quest for immanence ends in transcendence, but an abstract transcendence more intolerable than the concrete transcendence of religion. Thus [art, religion, and science]... are attempts... to reach... organized individuality.... Art comes nearest to success....

This conclusion of Collingwood is very impressive, but it leaves us with the problem open. And the problem, how to have unity and diversity combined, hinges on the problem of rationality. And, anyway, the problem of rationality is significant in its own right.

Anyway you look at it, there is still a significant and challenging differ­ence between the choice of the madman and the choice of the reasonable man; only the reasonable man has a number of reasonable alternatives, some of them equally reasonable, some more than others. This should be the desired corollary to any doctrine of rationality that may claim su­periority over the traditional doctrines of rationality.

VI. A THEORY OF RATIONAL DISAGREEMENT

Einstein agreed with the majority of physicists that quantum theory satis­fied certain desiderata. His contention, however, was that these desiderata are not sufficient. In other words, we need not specify a priori the de­siderata of rationality, but leave them open to rational debate. Also, given the desiderata, we may leave open to rational debate the question of which alternative satisfies them best, and if few are adequate, leave them to arbitrary choice without declaring this arbitrariness irrational.

This seems straightforward enough, and it applies to views as well as to conduct.

There are also two moves that seem widely acceptable (at least they have been proposed by Kant, and were endorsed by Neurath, Carnap, and practically all members of the Vienna Circle at one time or another), but which within justificationism do nothing to relieve the difficulty; within the above framework of rational debate are quite helpful. They are the idea of presenting universality as a methodological desideratum rather than as an ontological claim and the replacement of objectivity by intersubjectivity.

Let me explain.

It is agreed by practically all philosophers and historians of science that something similar to Bacon's pyramid of levels of generality takes place in science; the classical instances are Newton’s unification of Kepler’s and Galileo’s theory, electrodynamics as a unification of theories of electricity and magneticism, and electromagnetic field theory as a unification of electrodynamics and physical optics. The question is, what is the status of this pyramid? Does the pyramid in science correspond to a factual one, to one in the real world? Or is the pyramid something to which there can be no correspondence, is it a characteristic similar to the linguistic struc­ture of science? For justificationists this question makes little difference: they have to endorse each law of science, independently of the question on which level of the pyramid it happens to be. It is a secondary matter to them that laws are ordered in a pyramid, the primary matter being that the laws are true. This, however, raises an old difficulty: does not science rest on the a priori assumption that the pyramid does exist, that the next level of generality will be discovered, that laws will fit in a pyramid? In reply to this it might be claimed that whether^or not the pyramid reflects the true state of affairs in the cosmos, science may try to find it, and if the pyramid is there science may even be successful. Thus, the pyramid belongs to methodology. For justificationists, who claim that science proves its doctrines, this move is of no avail. For, when we prove a higher level theory we show a fortiori that one more level of the pyramid is filled by true laws. If we replace proof with probability, (meaning when we say that a scientific theory is probable, that most scientific doctrines are true) then the previous sentence will have to be modified, but not in a way prejudicial to the present discussion. And so, justificationists must accept the pyramid not only a posteriori, but also as the a posteriori corollary to any future development of science. This means that it is a transcendental part of all science, that for science to be possible God must have created parallel pyramids, one in nature and one in the intellect or in language or in some human quality. This is not as primitive a thesis as Wittgenstein’s picture theory of every statement in language, but it is identical with Bacon’s claim that God has created Nature and the Intellect on a par, so as to make science possible.²⁴

If, however, we claim that the lower-level theories are often refuted, or simply clash with the theories they replace even when the former are not refuted, then we can deontologize or methodologize the claim for universality in science. Moreover, in such a case, tests and crucial experi­ments are the particularizing effect that keeps the universalizing* effect going, that keeps us dissatisfied with existing universal theories and pushes us towards further generalities.

To take an example. William Whewell claimed that Newtonian me­chanics cannot be further deduced from a more general law. Why? The tendency, he said, was to explain both existing laws and the as yet unex­plained facts, by newer laws which are thus more universal. But in the case of mechanics there were no such facts. Simple. But where do such facts come from? Whewell stressed that there are no uninterpreted facts, and that interpreted facts follow from the theories they verify. So there can be no theory-isolated facts to increase the generality of a theory! To my knowledge this simple criticism of Whewell has never been noticed. To forestall dismissal of Whewell as an isolated case, let me point out that on this point Whewell’s view - of explaining theories and isolated facts by higher-level theories - was endorsed by Duhem and Meyerson.²⁵

The reason for the general significance of the difficulty implicit in this view - namely, that isolated or uninterpreted facts are unavailable - is that more and more philosophers tend to share it. They increasingly tend to endorse the view that there are no uninterpreted or isolated obser­vations of fact. This means, for almost all philosophers of science, that every new observation of fact confirms some known theory, which is a bit incredible; and that a new observation of fact cannot refute a theory except if we first invent a new alternative to that theory, which is even more incredible. There is little doubt that any critical examination of the current theory of interpretation, which (adumbrated by Kant, Johannes Mueller and Helmholtz) is due to Whewell and Duhem, is no satisfactory substitute for the traditionally most popular (Baconian-Lockean) doctrine of science as based on uninterpreted data. But there is an alternative theory of observation.

Both traditional theories of observation derive from their originators’ views of the role of observation in theoretical science. Bacon viewed facts as the raw material which is somehow processed into theories; hence his insistence on raw data. Whewell viewed facts as verifications; hence his view of interpretations as predictions. Popper views new facts as refu­tations, and so his theory of interpretation allows for both predicted and counter-predicted observations, the latter constituting novelties relative to the theories which they refute.

The result of this new view of interpretation is the methodological aspect of scientific theory. A theory is designed for explanation, it is thus the refuted theory, especially one which had once served a given purpose, which cries for a replacement which may better fulfill the same purpose. And here Whewell’s doctrine is rectified, as well as Duhem’s - into Popper’s.

Thus, I have argued, methodologizing or deontologizing universality can be effective only in a non-justificationist system. The same may be said of replacing objectivity with intersubjectivity.

Indeed, Kant replaced objectivity by intersubjectivity in order to deon- tologize universality, as should be obvious upon a simple reflection. But his deontology allowed no diversity, nor was it intended to. Nor did it remove the polarization of all human thought and action into the rational and the irrational. When Kant argued for the limits of reason and con­cluded that beyond these limits scientific intersubjectivity breaks down and can only be replaced by arbitrariness, he plainly and assuredly con­cluded from this that attempts to transcend the limits of reason must be irrational. This attitude of Kant’s should have made him view rationality in the narrow fashion common to his time, and practice it with consistency uncommon at any time. To an extent he was, indeed, quite impatient: not in the least glad to suffer fools and quite willing to consider all op­ponents as fools. Yet, if there is one aspect of Kant’s conduct which was grossly out of character, it was his ambivalence towards differences of opinion. I do not mean his table-talk: light-hearted controversies, even serious yet admittedly preliminary controversies, were tolerated as a matter of course. But Kant was willing to admire even opposition to the last as a symptom of a keen mind concerned with truth. It is perhaps amazing that such a prolific writer, who was so willing and able to elabo­rate a point, was so brief on some most crucial points of controversy, to the effect that we still know too little about his change from Leibniz- ianism to his own, critical, version of Newtonianism. No doubt, ambi­valence often is the source of reticence - certainly not any lack of frank­ness in this case. It is also quite possible to see the intellectual reason for the ambivalence: Kant’s critical doctrine of intersubjectivity seems to be a liberating doctrine, an idea enabling us to divorce rationality from unanimity; but it fails to do that: rather than establish possible under­standing between individuals, it clinched understanding to the point that it became full consent.

The newer theory of the intersubjectivity of tests of theories likewise does not help diversify thought and depolarize rationality and irrationali­ty. For, tests are confirmatory and, we are told, the acceptance of the confirmed theory is declared imperative. The same idea does, however, work if tests are attempted criticisms; they are then aimed at particulars and when successful they diversify - they open new areas of choice which are free for all.

In concluding this paper, which is intended more to open up issues than to close them, I wish to state three points.

First, if, as I suggest, instead of laying down desiderata we allow different contenders to state their own desiderata and debate them, then this may lead to an infinite regress. Indeed a regress there is here, just as there is the regress from one explanation to a higher level one; both these regresses may be infinite, yet they are not futile - not necessarily so; each step in the regress may offer its own interests and challenges.

Second, the desiderata may be solutions (perhaps of given character­istics) to given problems. Popper’s orientation has always been towards problems, and in his post-war work he has stressed that problems are the starting-point. I hope I have introduced this idea of his in a somewhat integrated fashion. To mention one corollary: Popper speaks of the objec­tivity of problems and perhaps also of their relative significance; but he says nothing about the objective choice of a problem to study - sometimes he even suggests one must fall in love with a problem, which smacks (quite erroneously, of course) of subjectivism. Once we suggest that problems are chosen by more general criteria, which themselves are open to criticism, we render matters of choice of problems inter-subjective to some degree.

Finally, the matter of choice of problems seems to me to be the crux of the philosophy of the future. Problems abound; yet most of them are not interesting, or not important, at least in comparison between the expected investment in, and the expected return from, the forthcoming solution. But all this is essentially highly speculative and vague. (Mbssbauer was assured by his superior that he was wasting time on a useless problem, very much as E. O. Lawrence before him, and as Planck even before.) And so this may lead to honest disagreement and even some debate. More important, the sense of significance is thus brought into research in a new way. In the classical doctrine it was brought in in the abstract, in the sense that science is important and requires dedication. Here it is introduced particularly, as a particular problem pertaining differently to any different research project. It thus leads anew to some connection between the prescriptive and the descriptive and again via rationality, but with a built- in diversity.