THE THREE VIEWS CONCERNING HUMAN KNOWLEDGE REVISITED

I apologize for my pedantry in advance. I shall put these matters in a table of 3 x 3:1 shall have to present the combination of instrumentalism and conventionalism under one label - I call it the active view.

I hope I am allowed the indulgence of explaining my choice of termino­logy.

(Historically it has been tacitly assumed that the problems of epistemo­logy and methodology were intertwined in that a solution to one is almost eo ipso a solution to the other: if we know how we learn, then we know that we know something; and if we know how we know, then we know that we have learned something. Consequently, when studies in one field proved too rough, it looked more promising to switch to the other field. All to no avail, of course; except that the unexplained shifts were con­fusing and the assumptions behind them tacit. One tacit assumption was, of course, that learning means the acquisition of some knowledge, just as improving one’s financial standing is the acquisition of some cash or credit. Now, no doubt, in coming closer to his target the fund raiser has raised some funds; yet the traveller comes nearer to his destination with­out partially being there. That learning is the acquisition of items of knowledge need not be assumed; and gradualism is the rejection of this assumption.)

My terms ‘essentialism’ and ‘instrumentalism’ come from Popper’s ‘Three Views’, where they are used to designate views concerning the rationality or aim of science.

The main disruptiveness in the table has to do with the fact that ratio­nality may be short term or long term: the rationality of the inquisitive view is the same as that of the contemplative view only in the long run. The short-term rationality of the contemplative view is not the truth but approaching it. In his Logic of Scientific Discovery Popper denounces long-term ends and stresses the short-term one; in his ‘Three Views’ he does the opposite. Yet this is no real incongruity, it being only a shift in emphasis; such shifts are unavoidable with the change in the problem to be tackled or the situation in which it is tackled.

My term ‘passivism’ is new, and is not meant as an insult in any way whatever, as an examination of the passivist literature may show. Here, for example, is Bochenski’s description (op. cit., 19) of objectivity as reflected in Husserl’s phenomenological method of intuiting essences, but which also is “an essential constituent of Western scientific method”: “... the investigator... must exclude everything that comes from himself, from the subject, above all his own feelings, desires, personal attitudes, etc. What is required of him is a detached observation of the object.... The reseacher who acts in accord with this rule is a pure knowing es­sence, one who forgets himself completely... the rule of objectivity requires a contemplative attitude, i.e., the exclusion of utilitarian considerations...”

It is perhaps the most incredible tribute to Kant that, due to his activist apriorism, the literature has completely overlooked the fact that Descartes was an avowed and strict passivist. I owe this point to a most moving paper by Ben Scharfstein ([‘Dream’]), and I can only refer the reader to that paper.

Of all the other terms, only two are new to any extent, ‘gradualism’ and ‘approximationism’; I regret not remembering where I first met them.

So much for terminology. As to the table itself, I wish to draw attention to its incompleteness, which is the same as the incompleteness of Popper’s scheme in his ‘Three Views’. Pointing out the incompleteness of a given schema may help orientation: it is no criticism of the schema which is not claimed to be complete. It may be useful in some cases to try and complete a schema and effect all possible permutations, etc. (see I. C. Jarvie, [Revolution], Ch. 3). This, however, is beyond my present purpose. To the limitations of my scheme, then.

First, Kant does not fit here at all. For, he explicitly refrained from judging the truth of his system and yet his claim for universal validity for it as higher than a claim for pragmatic truth alone. And certainly he was not a fictionalist; though an activist, he had no room for trial and error. (See [Conjectures], and my [‘Unity⁹])

Second, William Whewell does not fit here either. He fits the contem­plative view of the aim of science and of its status; but for the method of science he chose the active view. This is why he is so alluring, I suppose. But the allure is uncritical to such a degree that one has to appeal to the success of Newtonianism to explain it.

Third and final exception, the majority of contemporary philosophers of science, the so-called inductive school of probability, may be viewed as an alternative modified essentialism - indeed, unlike Popper’s modified essentialism it cannot consistently be incorporated within a modified conventionalism: what Popper calls “a liberal Utopia”, ([Conjectures], 351) is what Carl L. Becker described as The Heavenly City of the 18th Century Philosopher, a collection of individuals governed by nothing but enlightened self-interest and free of all tradition, convention and prejudice. I think the only place where this rather charming and incredibly naive aspiration is still - inadvertently - preserved is in the literature on proba­bility and induction.

With these exceptions out of the way, let us glance at my table and see what it illustrates. The contemplative and the active view share the idea of scientific certitude. They differ as to the source of scientific knowledge: the contemplative view is that the source is nature, the active view is that the source is the human mind. Hence the contemplative view is the view that scientific growth is entirely a passive contemplation of nature which is the intuition of true essences (methodological essentialism). It is this passivity, shared by both the inductivist and the (pre-Kantian) apriorist, which allegedly guarantees the truth of our theories (epistemological essentialism).

Fictionalism or instrumentalism became taboo since the foundation of the Royal Society, on the authority of the fathers of modern physics and of Sir Francis Bacon. Yet, Descartes already was willing to fall back on fictionalism or instrumentalism if this proved the only way to claim scien­tific status (see Sabra, [Light], 23ff.), perhaps following even Galileo and Torricelli in this respect (see Rossi, [Bacon], 221). Even the arch-skeptic and popular Joseph Glanvill took the same view ([Vanity], 211-212). The Cartesians graciously ascribed to the Newtonian philosophy mathema­tical but not physical certitude (Koyre, [Newtonian], Ch. 3); and the Newtonians, led by Helmholtz (Duhem, [Aim], 99ff), in like manner viewed fields of force as a useful fiction. Poincare (Hypothesis], Ch. 4) and Duhem (op. cit., Pt. II, Ch. 5) developed the view of science as useful fictions or empty definitions - recommending the latter as the better alternative. Later Eddington thought fictions were inescapable but strictly private, and empty definition was objective science ([Nature], 319-321).

Poincare stressed the conventional aspect of scientific theory, its being a set of empty definitions. Duhem stressed the instrumental aspect of it, the use of these definitions for practical purposes. He had an ulterior motive here, since his conservatism was not Newtonian but Aristotelian and nourished by his Roman Catholicism; he was, therefore, proud, he said, to find as an ally in adopting instrumentalism a person like Mach (333).

And so, by a switch, views of science deteriorated. First, my science was certainty-by-nature and true, whereas and your science was, as a consola­tion prize, certainty-by-convention and merely useful. Soon, all science was certainty-by-convention and merely useful, and certainty-by-nature be­came the prerogative of obscure metaphysics utterly divorced of all rela­tion to science. (See Appendix to Chapter 8.) Men of science who view all science as instrumental and definitional find themselves with a vacuum when it comes to ideas on the nature of things.

The inquisitive approach has ancient ancestry, of course, in the Socratic philosophy. Yet it is the most recent approach to science, for which we are indebted to Popper. (See my Novelty'].) Its chief characteristic is the idea that a given scientific theory, say Newton’s mechanics, is a descrip­tion of nature, though not always the best we have, and even the best we have need not be the last word. Even those who claimed to hold the Socratic approach to science could not usually stick to that view because of a strange feature of the situation. On the one hand, as we contemplate the application of the inquisitive approach to science, we face at once the problem of demarcation of science: if scientific theory is not certain or quasi-certain, what differentiates a good theory from a bad one? It is thus not surprising that Popper has stressed the problem of demarcation and its significance throughout his philosophical career. On the other hand, the more successful he was in his application of the inquisitive approach to science, the less important this problem turned out to be; quite apart from the fact that Popper’s initial solution has turned out to be either inadequate or false. In retrospect we may perhaps say, one theory is pre­ferable to another when and only when it is a better approximation to the truth. One may object to this as a very narrow sense of preferability, as I have shown by counter-example. This may be remedied by the introduc­tion of a theory of degrees of refutability. Popper himself has found it necessary to modify his theory of degrees of refutability: and he did so by relativizing them against some background knowledge; this cannot be viewed as more than a progress report: we do not yet have a sufficient theory of background knowledge (see my ['Background']). I have tried to offer elsewhere (['Nature']) examples illustrating cases where barely testa­ble hypotheses are preferred to highly testable ones in some intuitive sense of ‘testable’ perfectly acceptable to Popper. But, I agree, the situation in its actual history or in the abstract still leaves too many questions not fully answered: I suppose we are all at sea, and may benefit from stopping to discuss old solutions to the problem of demarcation or offer new ones and ask, instead, is the problem pressing enough to make us continue investigating it?

But if we proceed, I suggest a much broader sweep and a set of prob­lems to be arranged first. And we may study the problems within Pop­per’s philosophy or in a comparative study. Take Popper’s philosophy first.

The status of science is still not clear in Popper’s theory of science. What status has a refuted theory? An untested refutable theory? A tested un­refuted theory? A rescued theory? All these are open matters, or ones viewed as open by Bartley and myself, especially in view of the fact that the major questions of traditional philosophy are answered by Popper without going into the question of status. This is the main point which my table illustrates.

We may speak of status, however, merely by contrasting Popper’s view with the other two schools replacing the status of certitude with that of tentativity. This excellent move, however, is clearly not a demarcation proper: obviously, all unrefuted synthetic propositions are tentative, be they scientific or not. Dealing with the problem of demarcation proper, we may do it more historically - interpreting the history of science with the help of the gradualist view of the rationale of science. Then some degree of approximation to the truth - of verisimilitude-may perhaps be defined to demarcate science proper, and, again, we have only progress­reports here. This may perhaps be done after the incorporation of additional desiderata - which means changing our view of the rationale of science, in its actual history or in the abstract. Be our choice here what it may, I feel, we may do better to explore at first only the desiderata of degrees of explanatory power - a modification of Leibniz’s view to in­clude refutations - and see if it cannot suffice. It should follow from a good theory of verisimilitude that Newton’s theory, though false, is more verisimilar than Kepler’s. I suppose this will not suffice, but I cannot say. If it will not, then, obviously, our view of either rationality or methodo­logy will have to be altered.

Popper’s attitude here is still not clear to me. In his early work he repeatedly claimed that high degree of refutability is the only desideratum we need insist on. Not that he was unaware of other desiderata, like expla­natory power, generality, content, simplicity, but that, presumably, he viewed them all as taken care of by refutability. He says, for example ([Conjectures], 135n), “The reason why I consider the argumentative and the explanatory functions as identical... are derived from a logical ana­lysis”. In other cases he claims not identity but weaker kinds of reduci- bility, such as monotony. For example, he distinctly presents explanatory power, content, and confirmability as different; but he derives from his definitions of these functions that they increase together with the increase of improbability. I find all this of great heuristic value, but as satisfactorily refuted by a number of instances I have provided ([Nature⁹]). (I need not say the ‘logical analysis’ Popper speaks of is erroneous; indeed the error is not difficult to spot: Popper’s analysis is of a partially ordered set, and rests on ordering as if it were not partial. A theory which entails another both explains more and is more refutable; Weyl’s theory explains both Einstein’s and Maxwell’s, and is at least as refutable as either. The Bohr- Kramers-Slater theory was refuted without having explained a single phenomenon, much as Thales theory was. A highly ad hoc explanation, on the other hand, explains what it was set to explain and is quite possibly exhausted thereby, and is thus irrefutable. We may answer this last ponit by claiming, the explicanda are refutable, and so their explicans is refu­table too. Taking this reply seriously we will conclude that no ad hoc theories are possible, only partial ad hocntss exists. No doubt, then, Weyl’s theory is (partially) ad hoc in this sense and hardly refutable, yet highly explanatory all the same; whereas Einstein’s theory is less explana­tory and also less ad hoc. Whichever way you look at it, Popper’s alleged logical analysis is shoddy: even great analytic minds may falter on a rare occasion.) I consider the desiderata mentioned here quite significant, and perhaps the inquisitive view as described in my third column is already obsolete.

But there are other desiderata to be considered as well, and they may all define other kinds of rationality plus status; credibility, if one likes (I do not), usefulness, challenge, depth (this I like very much).

As to the ideas of challenge and of usefulness, they have not yet been attacked, and we do not know what incentive they constitute for what action, what is conductive to challengeability, or to use, under what con­ditions a challenge is taken, especially of possible usefulness of an innova­tion, etc. It is amazing how many obviously challenging problems are as yet unstudied. As to the idea of depth, there are hints about it in the literature, but nothing substantive or systematic. Popper claims depth to be undefinable. In a trivial sense this is true of all concepts; yet we can always try. I see no reason why we should leave this magnificent concept, and the allied one of enlightenment, unexplored. Popper’s view of credi­bility is not always very clear. He says, ([Logic], 414-5), “...while it is a mistake to think that probability may be interpreted as a measure of the rationality of our beliefs,... degree of corroboration may be so inter­preted... It must not be interpreted... as a degree of the rationality of our belief in the truth ofh; indeed we know that... [the degree of corrobora­tion of h is zero] whenever h is logically true. Rather it is a measure of the rationality of accepting, tentatively, a problematic guess...” I do not know what this may possibly mean. If when saying ‘accepting’ Popper means ‘apply’, then, as Popper himself says in his ‘Three Views’ ([Conjec­tures], 111-113), we do not hesitate to apply a theory which is not ‘a problematic guess’, but which we know to be false. If when saying ‘ac­cepting’ he means ‘believe’, then I do not see why belief should be rational in this sense: Popper himself confesses (e.g. [Logic], 38) his holding metaphysical beliefs which are, of course, uncorroborable, yet held quite rationally, of course. So, what on earth does he mean by ‘accept’?

Anyway, his theory of corroboration is supposed to have a wider scope than mere acceptability. I wish, however, to repeat that it is difficult to place the role of empirical backing or corroboration - in Popper’s sense of failed refutation - here (see my [‘Corroboration’]). This is particularly so since a scientific theory may be corroborated scientifically - through a genuine test - or not, just as an unscientific theory may be tested scienti­fically. For example, a superstituon may be tested and refuted or corro­borated scientifically. So can a theory not of any explanatory yet of tech­nological import, - a theory solving a technological not a scientific problem - be corroborated scientifically or not (we have scientific technology, pre, pseudo, etc.). For the sake of, and in the name of, social responsibility, a theory must be corroborated before it be institutionally implemented, as I have argued elsewhere (['Confusion⁹, II]). Viewing scientific theory as a contemplative institution - textbook - or as technology of sorts - educa­tional or otherwise - may resolve some of the difficulty. The difficulty, to repeat, is that the institutions fostering science are of a very different type from the institutional, or accepted, scientific ‘received’ opinion. One of the interesting questions which are today on the agenda of quite a few stu­dents of the topic (see my ['Kuhn⁹]) is, does the changing textbook of science through the ages cover the field of the history of science? Suppose the answer is affirmative. Query: does the science text­book only endorse corroborated theories? Suppose the answer is, yes. I will then have to modify my view of the role of corroboration in Pop­per’s philosophy of science, alter my view to one between my present one and Popper’s present one, and cease to see Popper’s theory as strictly a theory of conjectures and refutations. This, of course, purely on the descriptive level of discourse - studying science in its traditions. As to the prescriptive side, I may, and under the pressure of the theory of the impor­tance of corroborations in science I would, then, call for a reform of the constitutions of science, in the hope of its moving further towards a more open and free inquiry, towards a more problem-oriented and problem­solving attitude, where solving a problem interestingly will merit highly, regardless of whether the new solution be refuted in the first test or in the second.

How shall we go about this difference in viewing science and its tradi­tion? Do we decide to reform our view of the tradition or do we decide to reform the tradition itself? The first move in the direction of answering this question may well be the move in the direction of explaining corro­borations, and in particular of their past importance in the scientific tradition. It is easy to explain their importance in technology (see my ['Confusion⁹, II]). It is possible to show that corroborating evidence is at times important in science, though not as such. In reply to my claim that in science corroboration as such does not signify, Popper claims that historically corroboration meant the difference between the survival or perishing of modern science. I do not deny that. The novas of the late sixteenth century have played a similar role, yet we do not believe in them: we try to explain them. Suppose we explain the role corroboration played in the rise of science as the result of its competition with religion as a justi­fication. In this case we may see in Popper the last vestige of justificationism. Suppose, however, that without corroboration science as we know it may perish; at least as an institution. Why? Will this lead to a new intellectual phenomenon superior to science? Or will it be a disaster? If a measure of greatness of a philosophy can be given, it should be its hinting beyond itself. And I find Popper’s philosophy admirable in this respect.