1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar

The Ludwig von Mises Institute

Advancing Austrian Economics, Liberty, and Peace

Advancing the scholarship of liberty in the tradition of the Austrian School

Search Mises.org

What Empiricism Can't Tell Us, and Rationalism Can

Mises Daily: Thursday, January 26, 2006 by


It is no longer fashionable in political science to refer to "self-evident principles." Indeed, any reference to self-evident or axiomatic propositions is taken to be evidence that a scholar is leaving the realm of science and entering a mysterious netherworld consisting of tautologies, definitions, and metaphysical statements.


Thus, the great methodological debates within political science from early in the twentieth century to the present day represent what might best be described as "in-house" debates between scholars who take as given that political science must be an inductive and empirical discipline.

For example, the overthrow of the then dominant behaviorist/positivist epistemology and methodology in the middle third of the twentieth century was not precipitated by a debate about whether man should be studied empirically. Instead, the overthrow of behaviorism and positivism was precipitated by criticisms of the behaviorist epistemology and methodology from other empirically-minded scholars.

As another example we might cite the current quantitative/qualitative debate in political science. This debate is completely "in-house," in that all of the parties represented in the debate take as given that man must be studied empirically.[2] The debate is only about which empirical methods are best suited to political science.[3]

Given this broad consensus about the ultimate empiricist foundations of political science, it is perhaps appropriate to characterize the discipline as possessing a dominant "research tradition" in the sense that Larry Laudan employs the phrase.[4] In this sense we can describe the empiricist research tradition in political science as constraining virtually all of the debate about the proper questions to be asked in political science and the methods appropriate to the discipline.[5]

Even though the specific questions asked and methodologies employed within the discipline have evolved over time (such as the evolution from behaviorism to post-behaviorism and the eventual evolution from pluralism to rational choice theory, etc.) there has been no alteration in the basic underlying epistemological assumption that political science must be an empirical discipline. If this characterization of political science possessing a dominant research tradition is accepted, then an obvious question presents itself: Is the empiricist research tradition the one we ought to adopt?

I want to suggest that the empiricist epistemological position is not the one that we ought to accept, and that this epistemological position will make it literally impossible for political science to progress.


According to the empiricist, nothing about the social world can be known with apodictic certainty. Instead, social phenomena can only be tentatively known to be true after examining the social world. If you think you know something to be true about the social world you have to test your theory against real experience. While there are obvious differences between the methodological schools within political science in the way in which they apply this empiricist epistemological position to concrete problems, virtually every modern methodological school assumes that you have to observe empirical phenomena before you can draw any conclusions about the social world. In sum, the scientific enterprise looks something like this for empiricists for both natural and social science:



(Or, for pragmatists, a more "useful" theory)

Quite obviously, then, progress occurs in the social sciences in a manner exactly analogous to progress in the natural sciences.[6] Hypotheses are continually tested against new empirical data and are tentatively accepted as "true," or are rejected based upon the "evidence."

The cumulative sum of all of this empirical research gives us a certain amount of confidence in the theories that survive, although we are constantly attempting to formulate ever more precise empirically testable hypotheses.[7] One version of the empiricist epistemological position in particular took this idea of testing and empirical progress through accumulation to its logical end point.

This is, of course, the logical positivist epistemological position. The logical positivists claim that all non-testable propositions are either definitions, tautologies, or altogether meaningless.[8] If a proposition cannot be "verified" it is, ipso facto, meaningless according to the traditional logical positivists.

Prediction is thus an integral, if not the integral, part of the empiricist's idea of progress, (particularly for the logical positivists). For if it is true that we have to test our hypotheses against empirical evidence, the predictive ability of an hypothesis or theory quite literally determines its tentative acceptability.[9] If a theory or an hypothesis is incapable of predicting certain observable events that it purports to explain, then it is clear that even if the theory or hypothesis survives today it is destined to be at least partially refuted by future events.

We can see the unrivaled importance of prediction for the empiricist epitomized in the pragmatist philosophy of science. Although it may not be immediately obvious that pragmatists are empiricists, this fact emerges quite clearly once we recall that a pragmatist evaluates the utility of a theory only a posteriori; that is, there is for the pragmatist no way to establish the usefulness of a theory prior to its employment for some definite task.[10] While the pragmatist need not, and indeed rarely does, hold that a theory which predicts better than a rival theory is ipso facto "better," he does hold that a theory which better predicts those things which are immediately relevant to a specific problem is superior to a rival. The importance of prediction to the empirical political scientist can also be observed in the many critiques of the Rational Choice School. For the Rational Choice School is frequently berated for its inability to predict even general events, which is interpreted as evidence that the Rational Choice School is observably deficient in some respect.[11]


To claim that the discipline of political science is dominated by the empiricist epistemology and methodology is to state nothing revolutionary. Indeed, this fact would appear completely unremarkable were it not for the fact that there still exists an alternative epistemology and methodology; namely, Rationalism. The rationalist epistemology starts from the assumption that man can know at least some things about his world with absolute certainty, and without "testing" to see if they are true through experience.[12] Some of this knowledge can be acquired through mere ratiocination alone (the analytic a priori), while other knowledge must be acquired with a certain admixture of experience (the synthetic a priori).

For the social sciences, the rationalists claim, the synthetic a priori is vital. Because man can reason, choose, and act, it is imperative that we acquire knowledge about human action that is, while synthetic, necessarily true.[13] We must begin our study of the social world with this a priori foundation because man is not governed by time-invariant laws in the same way that we presume natural phenomena are.[14] Man, in short, can choose to act in one way today, but he may choose to act in the opposite way tomorrow. This epistemological position implies, quite obviously, that the methods to be employed in the study of man must be of a radically different nature than those employed in the natural sciences.

The primary reason we must rely upon the synthetic a priori in the social sciences, so the rationalist contend, is that without some sort of irrefutable axiomatic foundation for social science we have absolutely no way to know whether or not we are falling prey to the post hoc ergo propter hoc fallacy. In other words, there is no way to tell whether or not the "causal-nomological" patterns we empirically observe in the social world are "caused" by the things we think they are, or whether they just coincidentally related and have no necessary connection.[15]

There is, however, an even more powerful rationalist criticism of the empiricist epistemology. This is, quite simply, that every formulation of the empiricist epistemological and methodological position itself must be formulated in synthetic a priori terms.[16] For instance, the pragmatist epistemology, (as developed, for example, by Laudan), must be formulated in terms that are unmistakably a priori. To state that "science is problem solving" is to state a synthetic a priori proposition which purports to be true![17] Or again, to state that "thinking is an instrument of action" (as Dewey has done) is to state a synthetic a priori proposition. There is obviously no way to demonstrate the pragmatic utility of these propositions utilizing the pragmatic method itself.

As yet another example, from the pluralist/relativist camp, take Feyerabend's pronouncement that "[A]narchism, while perhaps not the most attractive political philosophy, is certainly excellent medicine for epistemology, and for the philosophy of science."[18] Would it be impertinent for us to inquire of Feyerabend whether he intends for us to assume that this proposition is necessarily and universally true? If he doesn't think this proposition is necessarily true then why should anyone read his book? Of course all of these propositions claim universal validity;[19] and if they claim universal validity then they are synthetic a priori propositions.

As yet another example, the hermeneutical epistemological position can similarly only be formulated solely in synthetic a priori terms. Hermeneuticians need not deny the existence of synthetic a priori knowledge,[20] but if they do, then they would have to state something to the effect that "social science can only proceed through the exegesis of written or acted texts." In that case, they would be stating a synthetic a priori proposition that claims universal validity. More frequently, the hermeneutician argues something to the effect that there are "multiple complimentary truths about a complex practice or text," and thus, "we never arrive at one absolute truth."[21] It hardly even need be said that these propositions are synthetic a priori propositions as well, which makes the content of the propositions baldly self-contradictory.

As a final example, let us examine the pronouncements of the logical positivists against the existence of true a priori propositions — or any metaphysical statements at all.[22] What has been said above equally applies to the logical positivists; namely, the anti-metaphysical, anti-a priori pronouncements of Carnap, Gödel, Neurath, Karl Menger, and the other Viennese positivists categorically denying the existence of synthetic a priori truths were all themselves synthetic a priori propositions.[23]

In sum, then, it is simply impossible to formulate a proposition concerning the existence of the synthetic a priori that is not itself a synthetic a priori statement.[24] As we shall now see, this truth has radically important implications for the related concept of progress in the social sciences.


The previous section of this paper attempted to demonstrate some of the inherent inner contradictions of the empiricist epistemological position in political science in all its various modern manifestations. It was argued that (1) it is impossible to determine solely on empirical grounds whether or not empirically-derived propositions are "true" or whether they are instantiations of the post hoc ergo propter hoc fallacy, and (2) that, as Johnson has argued, it is logically impossible to formulate a denial of the existence of synthetic a priori propositions that is not in itself a synthetic a priori proposition. What implications does this have for the idea of progress in political science?

In the first place, recognition of the fact that synthetic a priori propositions do indeed exist should spur the political scientist to go out and find some of them! What possible reason could there be for a political scientist to remain in the necessarily hypothetical realm of empirical research if there is a method through which we can acquire necessary knowledge about human action? For political scientists to remain solely in the empiricists' hypothetical realm is to act very much like geometricians would if they refused to utilize any geometric axioms. Can anyone possibly believe that any sort of progress could be made in geometry if every axiom was subject to empirical testing?

In such a situation, where no axioms were regarded as universally true, at least some geometricians might even be running around "testing" to see if every point on a circle was the same distance from the center! And why shouldn't they, if nothing can be known to be true a priori? The obvious reason that geometry can progress is that practitioners are able to deduce necessarily true propositions from the synthetic a priori axioms that form geometry's foundation. If such an axiomatic-deductive procedure were available for political scientists, wouldn't this methodology be superior to the empiricist's aimless and literally boundless search for what the empiricist himself admits to be only hypothetically true theories?

Of course, this defense of the synthetic a priori could be considered just so much empty verbiage if no axioms have yet been discovered that hold true with regard to all human action. Indeed, some empiricists might want to insist that they are open to the possibility of discovering such propositions, but until any have been found we must rely upon the only thing left to us — empirical investigation.

For the benefit of these potential converts to rationalism, let me advance some of the axioms of social science that have already been discovered, axioms that afford the political scientist an apodictically true foundation for erecting a deductive science of politics.[25] The essential axiom for social science, and the one upon which the entire superstructure of each of the social science disciplines rests, is Ludwig von Mises's "Man acts." This proposition cannot even be thought to be false, since any attempt to disprove it (even solely in one's head) would constitute an action in itself. The axiom itself is apodictically true, and it implies other axiomatic propositions about human action such as:

"Human action is an actor's purposeful pursuit of valued ends with scarce means. No one can purposefully not act. Every action is aimed at improving the actor's subjective well-being above what it otherwise would have been… Interpersonal conflict is possible only if and insofar as things are scarce… No form of taxation can be uniform (equal), but every taxation involves the creation of two distinct and unequal classes of taxpayers versus taxreceivers-consumers."[26]

  Bastiat makes sense: $30
These propositions are ultimately derived from the axiom of action, and are thus irrefutable without self-contradiction in the same way as the axiom of action. Recognizing that these propositions are axiomatically true, we can immediately see that progress for the rationalist means continually attempting to deduce propositions such as these from other prior axioms already known to be irrefutably true. There is no obsession with prediction or testing. The propositions are necessarily true — no testing is needed or is even possible.[27] The goal for the rationalist is to understand human action, not to predict something that is by its very nature unpredictable.

But then we come to the vital truth about political science once we come to recognize that progress in the discipline means improving our understanding of human action rather than improving our predictive ability as the empiricist claims. This is, quite simply, that the proposition "understanding human action is the goal of social science" is itself a synthetic a priori proposition that cannot be refuted without self-contradiction as well!

For any attempted refutation of this proposition would itself be a synthetic a priori proposition that purported to clarify our understanding of human action (specifically, our understanding of human action as it is manifested in social science itself); and the attempted refutation could not, and never could be, an empirical statement whose goal was to better predict human action.

The rationalist idea that man's goal in political science (like every science of human action) is to better understand human action thus cannot be refuted without self-contradiction! In this way, the rationalist epistemological position with regard to political science is totally vindicated. It is about time these irrefutable epistemological truths came to be accepted within the discipline, and the empiricist epistemological and methodological position rejected as offering no hope for political scientists.

Mark Crovelli is a graduate student in political science at the University of Colorado at Boulder. Send him mail. Comment on the blog.


Ayer, A. J. "The a Priori." In A Modern Introduction to Philosophy, edited by Arthur Pap Paul Edwards, 646-57. New York: Free Press, 1965.

Ayer, A. J., F. C. Copleston. "Logical Positivism — a Debate." In A Modern Introduction to Philosophy, edited by Paul Edwards, Arthur Pap, 726-55. New York: Free Press, 1965.

Blanshard, Brand. Reason and Analysis. Vol. 12, The Paul Carus Lecture Series. La Salle, Ill.: Open Court, 1964.

Brady, Henry E., and David Collier. Rethinking Social Inquiry: Diverse Tools, Shared Standards. Lanham, MD: Rowman & Littlefield, 2004.

Diesing, Paul. How Does Social Science Work? Reflections on Practice, Pitt Series in Policy and Institutional Studies. Pittsburgh, Pa.: University of Pittsburgh Press, 1991.

Ewing, A.C. "The 'a Priori' and the Empirical." In A Modern Introduction to Philosophy, edited by Arthur Pap Paul Edwards, 658-71. New York: Free Press, 1965.

Feyerabend, Paul. Against Method. revised ed. New York: Verso, 1988.

Gordon, David. "Hermeneutics Vs. Austrian Economics." In Ludwig von Mises Institute Working Papers Series. Auburn, Ala., 1986.

Hollis, Martin. Models of Man. New York: Cambridge University Press, 1977.

Hoppe, Hans-Hermann. Democracy: the God That Failed: The Economics and Politics of Monarchy, Democracy, and Natural Order. New Brunswick, [NJ]: Transaction Publishers, 2001.

———. Economic Science and the Austrian Method. Auburn, Ala.: Ludwig von Mises Institute, 1995.

———. The Economics and Ethics of Private Property: Studies in Political Economy and Philosophy. Boston: Kluwer Academic Publishers, 1993.

———. "In Defense of Extreme Rationalism: Thoughts on Donald Mccloskey's the Rhetoric of Economics." Review of Austrian Economics 3 (1989): 179-214. (PDF)

Johnson, Oliver A. "Denial of the Synthetic a Priori." Philosophy 35, no. 134 (1960): 255-64.

Kuhn, Thomas S. The Structure of Scientific Revolutions. 2nd ed. ed. Chicago: University of Chicago Press, 1970.

Laudan, Larry. Progress and Its Problems: Toward a Theory of Scientific Growth. Berkeley: University of California Press, 1977.

Popper, Karl R. "Unity of Method in the Natural and Social Sciences." In Philosophical Problems of the Social Sciences, edited by David Braybrooke, 32-41. New York: Macmillan, 1965.

Rothbard, Murray Newton. "In Defense of Extreme Apriorism." Southern Economic Journal 23, no. 3 (2002): 314-20. (PDF)

———. "In Defense of Extreme Apriorism." Southern Economic Journal 23, no. 3 (1957): 314-20.

———. Ludwig Von Mises: Scholar, Creator, Hero. Auburn, Ala.: Ludwig von Mises Institute, 1988.

———. "The Mantle of Science." In Scientism and Values, edited by Helmut Schoeck, James W. Wiggins, 159-80. Princeton, N.J.: D. Van Nostrand, 1960.

[1] Thus, we are sneeringly informed, the Austrian School of economics (which is both a school of economics and political philosophy) is a form both of "pseudo-science" and "propaganda." And why is this? Simply because the Austrians employ an aprioristic and deductive epistemology and methodology. See, Paul Diesing, How Does Social Science Work? (Pittsburg, P.A.: University of Pittsburg Press, 1991), p. 350. We are meant to assume, I suppose, that Diesing's own bilious and completely undefended comments about the Austrian School are not propagandistic? For similar unsubstantiated comments with regard to the Austrian School see, Mark Blaug, The Methodology of Economics (New York: Cambridge University Press, 1980), p. 93.

[2] Thus, Brady, Collier and Seawright's contribution to Henry E. Brady and David Collier, Rethinking Social Inquiry: Diverse Tools, Shared Standards (Lanham, MD: Rowman & Littlefield, 2004) contains the following strikingly illustrative comment: "Thus, we share [the quantitativist] view that quantitative and qualitative methods are founded on essentially similar epistemologies." p. 7.

[3] It is instructive to note that the so-called Perestoikan debate is also "in-house" in exactly the same way. The Perestroikans never question whether or not man ought to be studied empirically. On the contrary, they simply take issue with the fact that their own empirical methods are disparaged by other empiricists and are discriminated against in the journals and departments of political science.

[4] Larry Laudan, Progress and Its Problems: Toward a Theory of Scientific Growth (Berkeley: University of California Press, 1977).p. 79. I am, of course, not endorsing the rest of Laudan's pragmatic/relativist epistemological and methodological program.

[5] Ibid., p. 81. If what I claim is true, that there is an overarching empiricist research tradition, we can also set aside all the bickering about whether or not political science has a dominant "paradigm" as irrelevant. Even if there is not a dominant "paradigm" in the discipline there is still a shared research tradition under which all of the dominant methodologies can be subsumed; namely, empiricism.

[6] Of course, the logical positivists were the most vociferous defenders of this idea. See, for example, Karl R. Popper, "Unity of Method in the Natural and Social Sciences," in Philosophical Problems of the Social Sciences, ed. David Braybrooke (New York: Macmillan, 1965).

[7] Of course, the hermeneutical philosophers do not endorse this exact methodological prescription, but their own methodological pronouncements are nonetheless empiricist. The very idea of searching for meaning within written or acted "texts" necessarily implies that one does not know what's in those texts until one looks; i.e., this methodology is the same old a postiorism in different clothes.

[8] See, for instance, A.J. Ayer's debate with Father Copleston, where Ayer repeatedly denies to even understand Copleston's "metaphysical" statements. A.J. Ayer, F.C. Copleston, "Logical Positivism — a Debate," in A Modern Introduction to Philosophy, ed. Paul Edwards, Arthur Pap (New York: Free Press, 1965).

[9] It is interesting to note, however, that evidence not supportive of an hypothesis is not necessarily taken by an empirical researcher as "proof" that his hypothesis is wrong. In this respect, Kuhn's work is vital, because he was one of the first to recognize that empirical theories always have anomalies. (No empirical theory perfectly explains everything). See, Thomas S. Kuhn, The Structure of Scientific Revolutions, 2nd ed. ed. (Chicago: University of Chicago Press, 1970).

[10] Seen in this light, then, the relativist/pragmatist/pluralist "revolution" ushered in by Kuhn, Feyerabend, Laudan, et al., is really just a new formulation of the empiricist doctrine within the still dominant empiricist "research tradition." The very methodology of Kuhn, Feyerabend, Lakatos et al., (i.e., analyzing the actual history of science), is, while historicistic, still a postiori and empirical.

[11] This is not to say that the rational choice school is not an empirical school. On the contrary, this school is every bit as committed to the empiricist epistemology and methodology as the other schools. The members of this school also construct hypothetically true, probabilistic models of the social world which they then try to confirm a postiori through experience. Their methodology only appears to be deductive and "rationalist" to some because they attempt to logically deduce all the implications of their hypotheses. Their hypotheses, however, are no more axiomatic or a priori true than were the hypotheses of the Vienna Circle.

[12] On this see, inter alia, Brand Blanshard, Reason and Analysis, vol. 12, The Paul Carus Lecture Series (La Salle, Ill.: Open Court, 1964), A.C. Ewing, "The 'a Priori' and the Empirical," in A Modern Introduction to Philosophy, ed. Arthur Pap Paul Edwards, Free Press Textbooks in Philosophy (New York: Free Press, 1965), Hans-Hermann Hoppe, Economic Science and the Austrian Method (Auburn, Ala.: Ludwig von Mises Institute, 1995), Hans-Hermann Hoppe, "In Defense of Extreme Rationalism: Thoughts on Donald Mccloskey's the Rhetoric of Economics," Review of Austrian Economics 3 (1989), Oliver A. Johnson, "Denial of the Synthetic a Priori," Philosophy 35, no. 134 (1960), Murray Newton Rothbard, "In Defense of Extreme Apriorism," Southern Economic Journal 23, no. 3 (1957). Martin Hollis, Models of Man (New York: Cambridge University Press, 1977).

[13] Murray Newton Rothbard, "The Mantle of Science," in Scientism and Values, ed. Helmut Schoeck, James W. Wiggins (Princeton, N.J.: D. Va).

[14] I follow Hans-Hermann Hoppe here in referring to “time-invariant laws.” See, Hans-Hermann Hoppe, The Economics and Ethics of Private Property: Studies in Political Economy and Philosophy (Boston: Kluwer Academic Publishers, 1993).

[15] Hollis, op. cit. p. 66.

[16] The most succinct exposition of this idea has perhaps been made by O. A. Johnson in his aptly titled article “Denial of the Synthetic A Priori,” op cit. Also see the other references cited in note 10 above.

[17] Laudan, op cit., p. 11.

[18]Paul Feyerabend, Against Method, revised ed. (New York: Verso, 1988).

[19] If they did not claim universal validity then what would their epistemological status? Obviously, if they did not claim universal validity, they would then leave open the possibility that a priori knowledge exists.

[20] Indeed, Gordon has argued that Heidegger and Gadamer, for example, wrote nothing that can be interpreted as a denial of the possible existence of synthetic a priori propositions. David Gordon, "Hermeneutics Vs. Austrian Economics," in Ludwig von Mises Institute Working Papers Series (Auburn, Ala.: 1986).

[21]Paul Diesing, How Does Social Science Work? Reflections on Practice, Pitt Series in Policy and Institutional Studies. (Pittsburgh, Pa.: University of Pittsburgh Press, 1991).

[22] Ayer was perhaps the most vocal of the American logical positivists, and he was certainly not hesitant to deny the existence of the a priori. See, A. J. Ayer, "The a Priori," in A Modern Introduction to Philosophy, ed. Arthur Pap Paul Edwards (New York: Free Press, 1965).

[23] On this, see especially Blanshard, op. cit.

[24] Johnson, op. cit.

[25] I will apologize in advance to Diesing for what is to follow since he considers this foundation to be “rotten.” See note 1 above. It is terribly unfortunate for those of us who believe in the synthetic a priori that Diesing doesn’t share with us how he managed to disprove Mises’ rationalist epistemology. I can only hope that he will release this privileged information in the future so that I can amend my pseudo-scientific ways.

[26]Hans-Hermann Hoppe, Democracy--the God That Failed: The Economics and Politics of Monarchy, Democracy and Natural Order (New Brunswick, [NJ]: Transaction Publishers, 2001).

[27]Murray Newton Rothbard, Ludwig Von Mises: Scholar, Creator, Hero (Auburn, Ala.: Ludwig von Mises Institute, 1988). P. 44.