Claudio Gutierrez
I have reserved for this chapter the discussion of a position in economic methodology in relation to the interplay between assumptions and models that I consider especially important. It is the view that F. Machlup sets forth in his article on the problem of verification in economics. He presents in that paper an interpretation of what he calls the aprioristic and the ultra empiricist claims which, I think, is particularly worth mentioning. What the apriorist (Mises, Robbins) is trying to express, says he, in spite of their defective or inaccurate terminology, is only that fundamental assumptions of a theory cannot be subject to separate or independent verification. They can be rejected, of course, in the face of refuting evidence, but only as a whole and together with the system they belong in and which they contribute to define, and presumably on the condition that alternative systems are available. The ultra-empiricist, on his part, is only saying that all propositions of economics should be subject to independent verification. In other words, the ultra-empiricist, as against the apriorist, is asking that all investigation should start from facts. To this request the author answers: What facts? What is the criterion used to select, from the innumerable amount of facts which make up our environment, those privileged basic facts? He then proceeds, very aptly, to compare economic science to a machine that has several parts to it. Some of them are fixed–the body of the machine, the assumptions–and some of them are more or less movable–models of different degree of generality (MACHLUP 55).
Machlup's analysis seems to be basically correct. I think, however, that it can be completed with some reflections pertaining to general epistemology, and this is what I will attempt in the present chapter. I begin by recalling that according to the author (and I share his view) the two opposing claims, even if both are radical in their wording, are not both equally removed from the truth. As Machlup's witty remark –"what facts?"–, emphasizes, ultra-empiricism is radical not only in form but also in content. I do not think that one can consistently develop from it alone, without fundamental corrections, the complete framework of economic science as normally understood. It seems to be impossible, even by definition of what we understand by ultra-empiricism, to make it say, after a purge of radical vocabulary, that theoretical terms are after all essential to illuminate and guide any investigation. On the other hand, it does not seem to be a formidable task to modify the aprioristic claim so as to have it saying: "Fundamental assumptions are irremovable; other less fundamental propositions are, with a gradualism of sorts, variably removable". In spite of the radical sound of the aprioristic claim, this position can be adequately rephrased, as Machlup in fact does, so that one can develop out of it the normal system of economic thought. So, one will do well to take the aprioristic claim as basically sound, although inadequate in wording, and work on it to try to get the expansion or correction which may be needed.
For my part I think that two corrections to the claim should deserve special attention. The first is the addition of an explicit reference to the availability of alternative systems as a condition for the rejection of a set of fundamental assumptions; the second is the clear acceptance of a progression, of a more-or-less in the degree of independence of propositions. The latter seems to be conditio sine qua non for the plausibility of the methodological conception that we are describing: There must be a limiting kind of case in which fundamental propositions can be falsified through the independent test of some of the removable propositions. This is, of course, the grain of truth that the aprioristic claim saves from the sinking boat of empiricism. The former contains the basis for a possible justification of the self-validation or logical necessity of some assumptions which seems to be essential for the plausibility of the aprioristic approach itself. In other words, the position of Machlup is subject to attack from two different flanks. On the one hand, an apriorist will contend that it is not a matter of having pieces of machinery that cannot be removed, except with the system as a whole. The case is rather that it is not possible to remove certain intellectual "equipment" because it is necessary, that is, somehow identical with the mind itself. To remove those assumptions, if it is at all possible to call them by this word, would be to throw the baby away together with the bath water. On the other hand, the empiricist is going to say that one cannot save a particular hypothesis against adverse evidence by the easy expedient of changing some "less removable" pieces so as to be able to maintain both the hypothesis and the fundamental assumptions unscathed. As I think that both rejoinders are basically sound, I am going to try the correction of Machlup's approach along these two complementary lines.
Once we have taken sides with the aprioristic claim, however modified its version might be, it becomes clear that we have to accept a coherence theory of truth rather than a correspondence theory in the tradition of empiricism. Now, this understanding of truth is open to some objections against which, in contrast, the rival theory seems to be well protected. In particular, it is argued that coherence should not be taken as a criterion of truth, but only as a criterion of stability of belief, since coherence may equally well stabilize an erroneous or a true belief (POLANYI 64, p.284). NOTE 1 To this I have to answer that if we are going to seriously take into account the non formal roots of all knowledge we will have to admit that human imagination is limited and exhaustible in its creative power. We have to recognize two kinds of weaknesses–which may prove our greatest strength against the pitfalls of relativism–in the very person of the scientist. The first is the limitation implicit in the quantitative depletion of (available) paradigms, so that we could be forced to be content with the available-turned-necessary assumptions. The second is the limitation implicit in the qualitative depletion of a given paradigm, whose capability to defend itself against adverse evidence by theoretical maneuver may in fact become exhausted. I will presently attempt to give a more nearly full explanation of these two concepts. The important thing to note now is that, if the presence of the two weaknesses or limitations of the human mind is demonstrated, then what would only be "criterion of stability of belief" in the absence of those weaknesses, can be effectively proposed as a "criterion of truth," given them. For if the range of the beliefs that can be rationally entertained is somehow narrowed, a soundly stabilized belief could not but be accepted as the (humanly attainable) truth on the matter.
Frank H. Knight has a very unusual, and very illuminating, way of arguing the
case for the apriorism of economics. His point is that all knowledge is
empirical (in a rather extended meaning of the ordinary sense of the word) since
all propositions, even of logic and mathematics, are verifiable "to the extent
worth the cost by counting beans," and that the necessity of a priori
propositions comes not from an intuitive
power of ours, but rather from a
shortcoming of the mind, i.e., the lack of really creative imagination. We
just cannot think of an alternative world where such and such an
a priori law would not apply!
(KNIGHT 56, pp. 157-8)
Thus, if it might be true that any belief is somehow circular, and that all
beliefs tend to stabilize themselves, the lack of imagination inherently ours
can save us from having to rely completely on sheer subjective commitment. The
coherence of an explanation may be so wide in scope, so proximate in character,
or even so intimately appealing to our aesthetic or pragmatic sense, that the
possibility of furnishing a rival framework is exhausted or depleted for all
discernible effects.
What Knight is referring to in the above argument is, I should say, the quantitative depletion or exhaustion of paradigms, which, as has been shown, points to the rational end of knowledge, the circularity or self-validation of all theory. It corresponds, to borrow the phrase of T. Parsons, to the "valid precipitate" of idealism in the building of what we have called "transcendental realism." This exhaustion or depletion is found at the theoretical end of the problem of knowledge because it consists in the fact that our imagination is much too weak to conjure, out of pure air, enough paradigms to choose from in all circumstances. Its result is the necessity of accepting the one that we happen to be able to produce under the conditions. The necessary paradigmatic configuration need not be a single all-embracing one. As was shown previously, it most plausibly will be rather a set of two complementary frameworks, each of them having an exhaustion of its own and being unable by itself to give a full account of the whole of reality. This "internal" exhaustion, though, must not be confused with the one we are talking about here, the "external" depletion or exhaustion. The former is the exhaustion of the paradigm in its job of giving a consistent explanation of the phenomena of reality, whereas the latter is the exhaustion of the numerically available paradigms. NOTE 2
It is argued by Quine that "our statements about external reality face the tribunal of sense experience not individually but as a corporate body" (QUINE 60, p. xii). This means that all experience is systematic, in the sense that refutation of particular elements of a system of beliefs by adverse evidence can be absorbed by the paradigm by means of changing itself somewhere in the system. In this way, the system is able to still maintain the "refuted" element or hypothesis. Now, accepting as I do the coherence theory of truth implicit in this conception, I find myself unable to buy the extreme thesis that any hypothesis can be saved from refutation provided we make drastic enough changes somewhere else in the theoretical body. I rather think that the truth of the matter is that there exists, parallel to the quantitative exhaustion of paradigms, their qualitative exhaustion. This is a limit or depletion point which the paradigm reaches when it finds itself incapable of saving a particular hypothesis from empirical falsification. We can say that the paradigm reaches that point whenever the rest of the system loses its fluidity and becomes practically unmodifiable for the purposes of saving a particular hypothesis. The change to be made in order to maintain consistency is then qualitatively and not only quantitatively determined. The change has to be here, not there or at that other place. NOTE 3 Alternatively, we can say that the existence of this qualitative exhaustion point means that certain low-level interpretations of appearances cannot be thought away in any of the different versions of the paradigm or different articulations of the theory, and they tend, so to speak, to break away from the paradigm (Machlup's concept of independent verification). These low-level hypotheses or interpretations of appearances, and their rebellious conduct, are what furnishes the practical and theoretical basis for the existence of the empiricist point of view in our understanding of reality. The requirement of this positive depletion of the paradigm, complementary of its rational depletion, is–to speak again with Parsons–the "valid precipitate" of extreme empiricism for the building of our epistemological position. The fact that any paradigm should be subject to this point of empirical exhaustion is also interpretable as some weakness or fundamental limitation in our creative imagination. That limitation is the impossibility in which we find ourselves of thinking away the normal data of our sense experience and the results of the basic logical operations of inferences. NOTE 4
NOTE 1
Polanyi describes thus the armory of belief in defense of its own stability:
The intention
of the latter is different from that of the former two: While circularity and
epicyclicality "protect an existing system of beliefs against doubts arising
from any adverse piece of evidence, suppressed nucleation prevents the
germination of any alternative concepts on the basis of any such evidence" (pp.
288-91).
NOTE 2
An illustration of complementary paradigmatic explanations is provided by
Polanyi's description of the criticism of objectivism from a positivist point
of view and from a heuristic one: "Instead of indefinitely shifting an ever open
problem within the regress of the objectivist criticism of objectivist claims,
our reflections now move from an original state of intellectual hope to a
succession of equally hopeful positions. . . ." (p. 324). The regressive
criticism can undoubtedly lead us to something equivalent to the "hopeful state"
of a progressive heuristic posit, but only indirectly, so to speak, by showing
forth the fact of its own depletion. In its turn, the heuristic method is unable
to formulate a "bottom state" of pure empiricist intuition, although it somehow
points to an indirect or external relation to that basis of reality. One can
say that the alternative approaches are both important and necessary, especially
because they counteract the possible excesses of each other: danger of verbal
inflation or unwarranted speculation, on the one hand; danger of intellectual
depression, annulment of heuristic momentum, on the other. And although, as the
political economist will say, inflation is always preferable to depression, one
ought to try to have neither.
One can add that the two exhaustion points in
scientific paradigms may also be viewed as they themselves being complementary
paradigms, in the realm of methodology: the rationalist and the empirical
paradigms. This pair of paradigms have, of course, their own "external"
exhaustion point, so that the postulation of the pair is to be considered as a
necessary assumption.
NOTE 3 A most convincing argument for the existence of this exhaustion point was the central topic of a lecture given by A.Gruenbaum at The University of Chicago in the Fall of 1965, under the title "The Legacy of Pierre Duhem". He considered the content of the D-thesis to be that for every case of adverse evidence against a hypothesis H there exists a modified version of the theoretical context of H that "saves" H. He claimed the thesis to be unsound. Separate demonstration of the availability of "modified versions" is required in each case. Moreover, some counterexamples are producible in which the context of H is true beyond any reasonable doubt, at least with no lesser degree of certainty than that which a Duhemian would require to accept the falsification of the system as a whole. The counterexamples alluded to in the lecture were, I think, good illustrations of the loss of fluidity of the paradigm that defines the point of positive or empirical depletion.
NOTE 4 It is important to note that general first order logic, in particular if one does not include relations, is clearly identifiable as a low-level interpretation of reality, capable of breaking away from a more comprehensive logical paradigm. This, I think, explains the illusion entertained by early forms of logical positivism about logic being the very structure of reality, which was shattered by the dismal fate of the rest of logic in the face of repeated attempts to defining it in completely objective terms. A similar fate seems to be the one reserved for oversimplified dismissals of the problems in the foundations of mathematics, like Knight's principle of verification "by counting beans": what to do with those parts of logic and mathematics which do not have a decision procedure or even are demonstrably either-inconsistent-or- incomplete? I think that the only way out of this quandary is to interpret logic and mathematics in paradigmatic fashion, seeing their theories on a par with all other theory as being justifiable only heuristically. On the heuristic justification of mathematics, see POLANYI 64, pp. 124-131.