Mises and the Diminished A Priori
In a recent post, “Machlup and Mises,” on the blog Coordination Problem, Peter Boettke has called attention to and summarized an important paper, “The Epistemological Implications of Machlup’s Interpretation of Mises’s Methodology” written by Gabriel Zanotti and Nicolás Cachanosky. According to these authors, Murray Rothbard advanced an influential interpretation of Mises’s methodology that led mainstream economists to view Mises as an extremist.
Rothbard, Zanotti and Cachanosky claim, maintained “that Mises would have said that economic science is completely a priori, without any room for auxiliary hypotheses that are not directly deducible from praxeology” (p. 2). To understand this, we first need to consider what is meant by calling a propositiona priori . This is a proposition that can be known to be true just by thinking about it: you don’t need to examine the world to see whether it’s true. “2 + 2 = 4” is a priori true: once you understand what the proposition says, you can grasp that it’s true. You don’t need to keep counting objects to see whether the claimed equality holds true. By contrast, “Mises wrote Human Action” is not a priori true: just thinking about the proposition will not tell you whether it is true.
The view that economics is an a priori science struck many economists as absurd, and as a result Mises did not receive from the mainstream the attention that his work merited. Things might have been different, Zanotti and Cachanosky argue, had an alternative interpretation of Mises’s methodology prevailed over Rothbard’s extreme position. They acknowledge that Rothbard’s interpretation is plausible; but so, they say, is the alternative.
The alternative interpretation was advanced by a former student of Mises, the eminent economist Fritz Machlup; and it was in opposition to Machlup’s view that Rothbard defended his “extreme apriorism.” Machlup’s interpretation envisioned a much more modest role for a priori propositions.
According to Machlup, some parts of a scientific theory are not tested. These form the fundamental assumptions of the theory: without them, the theory would lose its identity. Newton’s physics offers a prime example “The whole system of physical mechanics rests on such fundamental assumptions: Newton’s three laws of motion are postulates or procedural rules for which no experimental verification is possible or required.” (p. 7, quoting Machlup)
In Machlup’s view, when Mises claims that economic theory is a priori, he has only this moderate position in mind. “The fundamental assumptions are neither logical nor factual.” (p. 16) (Later, Imre Lakatos called these assumptions the “hard core” of a theory.) In this interpretation, then, Mises did not make the extravagant claim that thinking without experimental verification discloses real truth about the world. Rather, to call a proposition a priori is simply to designate its place within a theory. We wind up with a domesticated Mises, fully in line with up-to-date “post-Lakatosian” philosophy of science. (When our authors say [p. 6] that Machlup makes a “Lakatosian turn” on the Duhem-Quine thesis, I trust they realize that most of Lakatos’s work was not published until the 1970s. Machup’s article was written in 1955.)
Zanotti and Cachanovsky’s analysis seems to me totally misguided. It proceeds through a false antithesis. Rothbard is alleged to claim that economics is completely a priori. The authors triumphantly cite passages from Mises that say he has introduced supplementary empirical postulates. “Mises says. E.g., The disutility of labor is not of a categorical and aprioristic character. We can without contradiction think of a world which labor does not cause uneasiness. ... But the real world is conditioned by the disutility of labor.” (p. 10, citing Mises) Have not our authors undermined Rothbard’s interpretation?
Not at all, and they know it. Later in their own article, they acknowledge that Rothbard was fully aware of the place of subsidiary postulates in Mises’s praxeology: “Rothbard (1957, p. 315) moves on to mention the role of empirical assumptions in praxeology: ‘Actually, despite the ‘extreme’ a priori’ label, praxeology contains one Fundamental Axiom — the axiom of action — which may be called a priori, and a few subsidiary postulates which are actually empirical.’” (p. 16) Why then do they say at the start that, according to Rothbard’s “extreme apriorism,” all of the subsidiary postulates are derived a priori?
Even if I am right that our authors present a distorted view of Rothbard’s interpretation of Mises, the decisive point in the controversy lies elsewhere. When Mises speaks of a priori knowledge in economics, what does he mean? To say that an a priori statement in a theory is one not subject to testing makes an incomplete claim. One needs also to ask, why is the statement immune from testing? One answer, that of Machlup, is that the statement is a mere convention: no claim is made for the truth of the statement in itself. Mises again and again makes clear that he does not look at matters in this way. He thinks that his a priori claims are incontrovertibly true, not just artifacts of a theory.
If our authors wish to dispute this, they need to cite passages from Mises which either state or imply that the a priori claims of praxeology are merely conventional. These passages would then have to be weighed against the many statements by Mises to the contrary. He says, e.g.,
The theorems attained by correct praxeological reasoning are not only perfectly certain and incontestable, like the correct mathematical theorems. They refer, moreover with the full rigidity of their apodictic certainty and incontestability to the reality of action as it appears in life and history. Praxeology conveys exact and precise knowledge of real things.
As if he were writing in advance to refute Machlup’s later interpretation of his position, he next says:
The starting point of praxeology is not a choice of axioms and a decision about methods of procedure, but reflection about the essence of action. There is no action in which the praxeological categories do not appear fully and perfectly. ... No experience can ever be had which would contradict these statements. Such an experience would be impossible in the first place for the reason that all experience concerning human is conditioned by the praxeological categories and becomes possible only through their application.
What, then, have Zanotti and Cachanosky to set against this? Nothing at all. They offer a number of quotations from Mises that to them show support for Machlup’s interpretation, but none of these addresses the key point: are the a priori claims of praxeology mere conventions or more than this?
I do not propose to go through all of their citations from Mises, but a few examples will show how they completely fail to see what is at issue. They cite Mises to the effect that praxeology does not convey a full knowledge of reality. Indeed it does not; Mises never claims to be able to deduce particular facts from praxeological theorems. What relevance has this for the status of a priori propositions in praxeology?
They quote this passage from Machlup: “‘Aprioristic reasoning is purely conceptual and deductive.It cannot produce anything else but tautologies and analytical judgments.’ While its [sic-this?] sounds like an ‘empiricist’s’ criticism of the aprioristic position, it is in fact a statement by Mises.” (p. 8) Machlup, and our authors following him, have made an unwarranted assumption. An analytic statement is true by the meaning of the concepts it contains. It does not follow from this that analytic statements are true by convention, i.e., that their truth arises from decisions about how to use words. That is a theory about analytic statements, not itself part of the meaning of “analytic statement.” Some of the logical positivists, and evidently Machlup as well, accepted this theory; and if you accept it, Machlup’s interpretation of Mises would be right. The a priori judgments of praxeology would be the products of convention.
But Mises rejects this theory. In the same passage quoted a few paragraphs ago, Mises says:
It is customary in the treatment of the epistemological problems of economics to adopt one of the solutions suggested for the natural sciences. Some authors recommend [Henri] Poincaré’s conventionalism. They regard the premises of economic reasoning as a matter of linguistic or postulational convention. ... However, the sciences of human action differ radically from the natural sciences. All authors eager to construct an epistemological system of the sciences of human action according to the pattern of the natural sciences err lamentably.
Our authors cite this in support of Machlup’s interpretation:
Economics does not follow the procedure of logic and mathematics. It does not present an integrated system of pure aprioristic ratiocination severed from any reference to reality. Inintroducing assumptions into its reasoning, it satisfies itself that the treatment of the assumptions concerned can render useful services for the comprehension of reality. It does not strictly separate in its treatises and monographs pure science from the application of its theorems to the solution of concrete historical and political problems.It adopts for the organized presentation of its results [a] form in which aprioristic theory and the interpretation of historical phenomena are intertwined. (p.11, Z and C’s emphasis)
They ignore what Mises says almost immediately after this.
There are no such things as a historical method of economics or a discipline of institutional economics. There is economics and there is economic history. The two must never be confused. All theories of economics are necessarily valid in every instance in which all of the assumptions presupposed are given. Of course, they have no practical significance in situations where these conditions are not established. The theorems referring to indirect exchange are not applicable to conditions where there is no indirect exchange. But this does not impair their validity.
Mises makes the essential point even clearer in a slightly earlier passage:
But the end of science is to know reality. It is not mental gymnastics or a logical pastime. Therefore praxeology restricts its inquiries to the study of acting under those conditions and presuppositions which are given in reality. It studies acting under unrealized and unrealizable conditions only from two points of view. It deals with states of affairs which, although not real in the present and past world, could possibly become real at some future date. And it examines unreal and unrealizable conditions if such an inquiry is needed for a satisfactory grasp of what is going on under the conditions present in reality.
However, this reference to experience does not impair the aprioristic character of praxeology and economics. Experience merely directs our curiosity toward certain problems and diverts it from other problems. It tells us what we should explore, but it does not tell us how we could proceed in our search for knowledge.
Amazingly, Zanotti and Cachanosky quote part of this very passage, as if it supported rather than refuted them.
Mises, contrary to the “post-Lakatosian” predilections of our authors, does not advocate an amalgam of theory and empirical inquiry in one discipline: he rejects this utterly. His point is rather that once we know empirically that a phenomenon is present, e.g., the existence of money, the deductive theorems about that phenomenon then apply. The fact that parts of praxeology apply only under certain conditions is irrelevant to the key issue between Rothbard and Machlup: the status of a priori knowledge in Mises’s economics.
Our authors fumble again when they say:
Disutility of labor is assumed to be present in either a barter or monetary economy. In turn, the presence of money is a more general assumption than the presence of either commodity money or fiat money. This puts Mises under the same methodological structure as Machlup. It should also be noted that Human Action is a treatise on economics and Machlup’s piece was a paper within a specific methodological debate with detailed examples. It is to be expected that the assumed conditions in a treatise of economics to be more general than the ones present in a paper like Machlup’s. The fact that none of this proves or disproves a general theory is consistent with Machlup’s philosophy of science, according to which there is no deductive proof but a humble non-disconfirmation — empirical data illustrates, rather than tests, a theory. (p. 14 [I have corrected a typographical error in the quotation])
They are right that empirical data illustrate rather than test a theory, but for Mises, deductive proof is available. Once we know money exists, then everything deduced from the concept of money is guaranteed to apply.
There is yet another suggestion in Zanotti and Cachanovsky’s paper that is seriously misleading. They regret that Mark Blaug and others adopted Rothbard’s interpretation of Mises over that of Machlup. Why do they think that these authors drew from Rothbard rather than from Mises himself? The strong view of a priori knowledge in economics that repelled Blaug is clearly present in Mises, not a mere controversial “interpretation” advanced by Rothbard. Mises’s claim that we have a priori knowledge of the actual world is a stumbling block to the positivists and a scandal to Lakatos; but an unsupported reinterpretation of this claim will not make it go away.
 All page references in the text are to the posting of Zanotti and Cachanovsky’s paper on the SSRN website, available for download here: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2229570&download=yes (The paper by Rothbard which they criticize is his “In Defense of ‘Extreme Apriorism,’” Southern Economic Journal, January 1957, pp. 314-320. http://mises.org/rothbard/extreme.pdf) Though the point does not affect their paper’s argument, it is worth noting that the authors here misapprehend what Mises means by “praxeology.” Hypotheses are not deduced from praxeology, which is the science of human action. Rather, praxeology consists of a set of truths, at least in part deduced from the concept of action. The place of auxiliary hypotheses will be discussed later.
 This discussion is oversimplified. Not all philosophers accept the a priori status of “2 + 2= 4.” Some philosophers do not think there are any a priori truths, and some think that mathematical truths are empirical.
 I am concerned here only with the authors’ presentation of Machlup’s ideas. I do not address the issue of whether they correctly state his position.
 Ludwig von Mises, Human Action, Scholar’s Edition Auburn, Ala.: Mises Institute 1998), p. 39.
 Ibid., pp. 39-40
 Ibid., pp. 66-67.