Mises and Popper had a lot of ideas in common, especially classical liberalism, and they were both founding members of the Mont Pelerin Society but they also had some differences and these prevented them from forming a strong partnership. One major difference was in epistemology and method, where Mises insisted that the methods of the natural sciences could not be used in economics. Another was in economic policy were Popper was happy with state intervention in the economy, though not nearly as much as some of his libertarian critics suggest.Francesco Di Iorio has written a fascinating paper to show that Popper and Mises are quite close in their epistemology and methods, and only minor modifications on each side are required to bring them into close alignment. On policy, I think Popper just needed to understand the way that monopolies and unemployment are largely due to bad forms of intervention. The kind of social engineering that he advocated would have revealed the failures of defective interventions long ago, if the strong interventionist had only been prepared to adopt the CR approach.
Picking up what Mises wrote about Popper and the methods of the natural sciences. Human Action (1949), page 32 of the 1966 edition. “The modern natural sciences owe their success to the method of observation and experiment. There is no doubt that empiricism and pragmatism are right as far as they merely describe the procedures of the natural sciences”.
In the light of CR that is a false statement.
And later “The positivistic principle of verifiability as rectified by Popper is unassailable as an epistemological principle of the natural sciences.” (1962)
Is that any better? How did the positivistic principle of verifiability become unassailable when rectified by Popper? That appears to depict the debate between Popper and the positivists as a difference on detail rather than a difference that went to the core of the positivistic program. Many people like the late Richard Rorty like to think that Popper’s dispute with Carnap et al was a “family matter” which had no implications for people outside the positivistic clan. As though the use of data for verification or testing was a mere detail or even a verbal quibble, a matter of the language of discourse rather than the conduct of science (that may have been Hempel’s position).
In a nutshell, Popper shifted the focus from the justification of beliefs to the formation of critical preferences for publicy articulated (objective) theories. However all the introductory books on phlilosophy that I have been scanning lately are still stuck with the justification of beliefs, so Popper is relegated to small section in the philosophy of science and is rarely even mentioned in the chapters on epistemology, metaphysics, philosophy of mind, logic, politics and the social sciences. Getting back to Mises.
In The Ultimate Foundation of Economic Science (1962) Mises wrote a short section on Confirmation and Refutatbility
In the natural sciences a theory can be maintained only if it is in agreement with experimentally established facts. This agreement was, up to a short time ago, considered as confirmation. Karl Popper, in 1935, in Logik und Forschung pointed out that facts cannot confirm a theory; they can only refute it. Hence a more correct formulation has to declare: A theory cannot be maintained if it is refuted by the data of experience. In this way experience restricts the scientist’s discretion in constructing theories. A hypothesis has to be dropped when experiments show that it is incompatible with the established facts of experience.
OK, unless you want to quibble about details.
It is obvious that all this cannot refer in any way to the problems of the sciences of human action. There are in this orbit no such things as experimentally established facts. All experience in this field is, as must be repeated again and again, historical experience, that is, experience of complex phenomena. Such an experience can never produce something having the logical character of what the natural sciences call “facts of experience.”
There are examples of patterns being repeated in history – price controls, tariff protection, inflating the money supply. Even in complex phenomena there are patterns that may be clear enough if you know what to look for.
If one accepts the terminology of logical positivism and especially also that of Popper, a theory or hypothesis is “unscientific” if in principle it cannot be refuted by experience. Consequently, all a priori theories, including mathematics and praxeology, are “unscientific.” This is merely a verbal quibble. No serious man wastes his time in discussing such a terminological question. Praxeology and economics will retain their paramount significance for human life and action however people may classify and describe them.
It is most unfortunate that the demarcation of Science became such a big issue post Newton and in the Vienna positivists. Why did Popper think that demarcation was one of the two fundamental problems in the philosophy of science? This is my gloss on the matter (taken from a recent comment on the blog ThinkMarkets):
It all depends what you mean by Science. Originally to be scientific merely meant to adopt a systematic, deliberate and maybe experimental approach, whether in academic topics, cooking or angling.
Post Newton science turned into Science and some kind of empiricist or inductive Scientific Method was enthroned as the gold standard. This prompted a kind of “cargo cult” with people doing their best to mimic the activities that were supposed to deliver Scientific “cargo”.
Hume put a spoke in the Inductivist wheel and the inductivists are still trying to save Science by finding some way to make Induction work. Kant invented apriorism to save Science from Hume and this prompted many diverse lines of thought including Misian praxeology which functions as an alternative to Inductivism by providing a method of discovery and proof without resorting to empiricism.
Popper took on board a modified form of the Kantian a priori (Barry Smith later called this “fallible apriorism”). He claimed that there is no such thing as the Scientific Method and he showed that Inductivism is not necessary for science (small s).
What if we stop agonizing over Science and science and just talk about the comparative merits of theories in terms of their capacities to solve problems (understanding and pattern prediction), to unify differents areas of discourse,to stand up to various kinds of tests, and to be helpful in practical matters. Newtonian physics did well on those criteria, even though it was not the end of the road for physics, and so does Austrian economics.
Back to Mises and his dismissal of Popper on the ground that the demarcation criterion is a verbal quibble. Of course the words do not matter and the criterion needs to be seen in the context of the debate with the positivists and also in the context of scientific practice. It is really a question of whether or not evidence can be used in discussing the truth and falsity of theory or a body of knowledge. If evidence is not applicable then according to the criterion (or definition) then the theory in question is not a part of empirical (testable) knowledge. Actually that is a tautology, if a theory is not testable it is not testable. But the practical point is to find out sooner rather than later whether the theory can be tested (bearing in mind that there are degrees of testability) and then to discuss how it can be tested. Remember that testing is only one of the five methods of criticism.
- The check on the problem. Does the theory solve the problem?
- The test of logic (internal consistency).
- Consistency with other well-tested theories.
- The test of evidence.
- The check on the metaphysics.
Of course we know that Mises and the other Austrians are just as interested in facts as anyone else, but how do they propose to use them? Pressing on with Mises.
The popular prestige that the natural sciences enjoy in our civilization is, of course, not founded upon the merely negative condition that their theorems have not been refuted. There is, apart from the outcome of laboratory experiments, the fact that the machines and all other implements constructed in accordance with the teachings of science run in the way anticipated on the ground of these teachings. The electricity-driven motors and engines provide a confirmation of the theories of electricity upon which their production and operation were founded. Sitting in a room that is lighted by electric bulbs, equipped with a telephone, cooled by an electric ran, and cleaned by a vacuum cleaner, the philosopher as well as the layman cannot help admitting that there may be something more in the theories of electricity than that up to now they have not been refuted by an experiment.
That line of argument is really not worthy of a serious scholar. To observe the instrumental value of theories does not to refute the theory of conjectural knowledge. The instrumental use of a theory does not represent a confirmation of the theory. The Ptolemaic system could be used for a lot of practical applications. It just means that the theory is near enough for practical purposes.
Moving on to David Gordon who wrote a paper called The philosophical origins of Austrian economics to describe how the Austrian school arose in opposition to the German Historical School and then had to counter logical positivism of the Vienna Circle. He noted that this achieved most of its influence after the members fled from Hitler and set up the movement that became known as logical empiricism in the US. I think of this as Hitler’s revenge. Gordon wrote that this is the major influence that prevents most American economists from accepting praxeology and Austrian economics.
Gordon, like Mises, regards Popper as a part of the problem and he quoted the passage above where Mises referred to Popper’s “verbal quibbles”. Gordon then pursues the following argument.
It is easy to see that Mises’ reaction to the verifiability criterion would be the same. Praxeology arrives at truth by deduction. If someone wishes to define “meaning” so that the conclusions of praxeology are empirically meaningless, why should he care? To this an obvious rejoinder suggests itself. The logical positivists did not view their criterion of meaning as an arbitrary proposal, to be dismissed by anyone not sharing the Circle’s affinities. On the contrary, they claimed that their position was well supported. Are they correct?
I do not think so. In point of fact, the criterion is worthless, since every statement comes out verifiable under it. Suppose that “p” is a non-controversially verifiable statement, e.g., “there is a chair in this room.” Let us take “q” to be a statement logical positivists reject as meaningless. A good example is one that Rudolf Carnap held up to ridicule when he called for an end to metaphysics. He cited the following from Martin Heidegger’s Being and Time (1927): “The not nothings itself.” I shall not attempt to explain this: one can see why Carnap presented it as a paradigm instance of a meaningless statement.
Does the verification principle eliminate it? Surprisingly, it does not. From p, we deduce p or q. (This step is non-controversial.) Assuming that a logical consequence of a verifiable proposition is itself verifiable, (p or q) is verifiable. Further, if p is verifiable, then the negation of p is verifiable; this principle seems difficult to question. Now, consider this argument:
p or q, not -p, then q
This argument is valid, and each of its premises is verifiable. Then, q is a logical consequence of verifiable propositions, and it, too, is verifiable. Clearly, if the verification criterion cannot eliminate “the not nothings itself,” it is not worth very much.
A falsification criterion fairs no better. If p is falsifiable, then (p and q) is falsifiable. Once more, not-p should be falsifiable if p is, though Karl Popper has implausibly denied this. By an argument parallel with that for verification, we conclude that q is falsifiable.
And to conclude
To my mind, the foregoing considerations dispose of logical positivism, at least for our purposes. Because of Karl Popper’s great influence on contemporary economic methodology, however, I think it advisable to make a few remarks about his variant of positivism.
Popper has had some effect on Austrian economics, in large part owing to the fact that Friedrich Hayek, his close friend, has to some extent abandoned praxeology and adopted falsificationism. In doing so, Hayek reemphasized a positivist strain in his thought which has been present since his university days. He has been deeply impressed by the physicist and philosopher Ernst Mach, whose views in many respects resembled logical positivism. Mach rejected concepts in physics which could not be derived from the senses. For example, he refused to accept Newton’s doctrine of absolute motion because in his opinion it lacked empirical reference. He also rejected atomism: atoms did not really exist but were a mere hypothesis.
Hayek’s Machian tendencies emerge in full force in The Sensory Order, his study of perception. Popper cannot be blamed or credited with Hayek’s positivism. What he did was to help bring about Hayek’s extension of positivism to economics.
But this has been a digression. To return to Popper, his basic doctrine modifies the verifiability criterion. Rather than say that a meaningful statement about the world must be empirically verifiable, Popper asserts that a scientific statement must be falsifiable. Popper utterly disclaims association with the positivists: he stresses that his falsification criterion is a test for scientific statements, not a criterion of meaning. At least in his earlier years, though, he set little store by non-scientific statements; and although he has in recent times grown increasingly willing to countenance “metaphysical” statements, he does not consider them true or false. Small wonder that Carnap and Herbert Feigl classed Popper as an ally.
To say that a proposition must be “falsifiable” instead of “verifiable” at first seems trivial. If a proposition is verified, its negation is falsified; if a proposition is falsified, its negation is verified. Consider, e.g., “The demand curve slopes downward and to the right.” Whenever this is verified, its negation, “the demand curve fails to slope downward and to the right” is falsified.
Further, since any proposition is verifiable (as shown above), the negation of any proposition is falsifiable. But a proposition’s negation is of course also a proposition. Its negation is then falsifiable. Since this negation is identical with the proposition from which we started, we conclude that any proposition is both verifiable and falsifiable.
What then is all the fuss about? Popper’s falsification criterion is in fact much more than a triviality. He maintains that confirming a proposition does not add to the probability that it is true, since he rejects induction. No matter how many times a demand curve has been found to slope downwards and to the right, the chances that this statement is true have not gone up. Mises displayed characteristic good sense in having nothing to do with Popper’s skepticism.
If I was not already acquanited with so many wrong-headed criticisms of falsificationism, then I wouldn’t believe anyone would make these arguments.
Gordon rightly explains that if P is verifable, thenP v Q is verifiable; he then claims that if P is verifable, then ~P is verifiable. The negation of a verifiable statement is not necessarily verifiable, but since Gordon argues that “this principle seems difficult to question,” a charitable interpretation of the claim is that by P he is referring to the statement “there is a chair in this room.” Assuming the room is sufficiently small and well lit, both P and ~P are verifiable; in contrast, if P referred to the statement “there is a chair in this solar system,” then ~P could not be verified. Moving on, Gordon asks us to consider this argument:
But why? The verified statement in the above argument is ~P, so the verificationist has no business including the formula P v Q among his premises, since without verifying P or Q independently, P vQ is unverified. If Gordon means to suggest that both P and ~P have been verified, then all that has been demonstrated is that anything follows from contradictory premises.
Gordon seems to believe that both P v Q and P can be introduced as premises merely because they are verifiable, but then his criticism could be simplified by demonstrating that the negation of any verifiable statement is also verifiable, and the elaborate chain of reasoning would be unnecessary. A verificationist, meanwhile, would maintain that only premises that have actually been verified (not those that are merely verifiable) are eligible, and among those there are no contradictions.
Although above I urged a charitable interpretation of Gordon, his later application of the same argument to falsificationism seems to betray a fundamental misunderstanding of logic. Once more, he states, without qualification, that “~P should be falsifiable if P is,” but the negation of a purely existential statement (i.e. a statement unbounded by an observable location, e.g. a small, well-lit room) is logically equivalent to a universal statement:~Ex[Fx] =||= Ax[~Fx]According to Gordon, if P refers to “there exists a white swan,” then the statement “there does not exist a white swan” is also verifiable, but the latter statement is logically equivalent to “all swans are not white.” Therefore, if Gordon’s argument were a valid “criticism” of either verificationism or falsificationism, then he has inadvertantly demonstrated that universal claims of scientific theories can indeed by verified by observation.
Gordon then rounds of his misunderstanding of Popper by failing to appreciate that falsifiability was put forward to deal with universal statements. If a statement is verified, then its negation is falsified (and vice versa), and so every falsification must entail a verification. However, Popper stressed the asymmetry of this relationship with regard to our limited experience: although every falsification entails a verification, only existential statements can be verified.
On a related note, I’ll be attending the Austrian Scholars Conference at the Mises Institute in Auburn, Alabama from March 11-13.
Lee,
Like you I’ve seen a lot of such elementary errors. After a while you start to wonder whatever possesses these people to imagine that they should publish. I mean .. yes .. criticism is great and all that, and everyone is fallible. But why doesn’t anyone bother to get this kind of elementary nonsense criticized and corrected before putting their name on it and blasting from the rooftops?
Ken,
I actually retract some of my comments above. I misinterpreted Gordon’s argument. His point was that positivists had assumed that “meaningfulness” would be transmitted from premises to conclusion like truth. His argument presents a counterexample to that assumption, where a “meaningless” conclusion is derived from “meaningful” premises. Although this seems a rather weak criticism of positivism, because P and ~P cannot be verified simultaneously, it is not a mistaken as such — Gordon’s presentation suggested a much stronger and interesting argument!
Gordon was also correct when stating that if P is “meaningful”, then so must be ~P. His use of the word “meaningful” was consistent with its use by many of the original positivists — or at least that is his claim. That said, Gordon’s “meaningfulness” appears ridiculously strawman-like, because every “meaningful” statement would trivially have “meaningless” consequences. For example, “there is a white swan in this room” entails “there is a white swan”, but the latter is supposedly meaningless because its negation cannot be verified.
However, the one part of Gordon’s critique that was wildly misleading is the application of his arguments to falsifiability. His point that it can no more function as a criterion of meaning than verifiability is accurate, but his suggestion that falsifiability was originally suggested to solve that problem borders on dishonesty. Popper (and other falsificationists) have repeatedly stated that falsifiability was not proposed as a criterion of meaning and can never function as such.
I hope that I may be allowed to respond at this late date, as I have only today encountered this blog.
In the passage that Lee Kelly cites, I was concerned only to ask what would happen if falsifiability were taken as a criterion of meaning. I did not say or imply that Popper originally suggested falsifiability as a criterion of meaning.
To the contrary, I expressly repudiate this idea, which Mr. Kelly is right to criticize If Mr. Kelly had taken the trouble to read a little further in my paper, he would have found this passage: “But this has been a digression. To return to Popper, his basic doctrine modifies the verifiability criterion. Rather than say that a meaningful statement about the world must be empirically verifiable, Popper asserts that a scientific statement must be falsifiable. Popper utterly disclaims association with the positivists: he stresses that his falsification criterion is a test for scientific statements, not a criterion of meaning.”
In view of this, I do not think it is altogether fair to say that I have made a wildly misleading statement that borders on dishonesty.
Though readers of my book reviews could with justice say that I am one to talk, I think that Mr. Kelly would be well advised to tone down his rhetoric.
David,
You’re right. I missed that passage in the first time; I retract that with the other objections. However, had I read that passage, I would have also taken issue with the following claim:
This is just not true. Popper spent most of The Open Universe, for example, arguing that indeterminism was true. He considered it desirable for statements to be scientific, because it tends to make them easier to criticise — it is surely necessary for any empirical investigation. But Popper never, to my knowledge, ever said that metaphysical statements were neither true nor false.
This is entirely right, and I am grateful to you for catching this mistake. I ought to have said that Popper thought that metaphysical statements cannot be proved to be either true or false.
Turning to another matter, what you said about the “ridiculously strawman-like” nature of my use of “meaningful” seems to me mistaken. Every meaningful statement does not have meaningless consequences, as you suggest. What I venture to suggest you fail fully to grasp is that the positivists changed the verifiability criterion. Under their revised criterion, (it’s not my strawman; it’s their criterion) it’s no longer the case that to pass the test a statement must be directly confirmable by observation. It’s enough that it be the negation of a statement so confirmable, or a logical consequence of statements of either of these kinds, and so on indefinitely.
Thus “there is a white swan” is not meaningless because its negation cannot be verified. The argument goes the other way: because the former statement is verifiable, the negation meets the criterion as well. To reiterate, the positivists are not making the false empirical claim that whenever there is a statement that is directly confirmable by sense observation, its negation will also be directly confirmable by sense observation. They are stipulating an extension of the criterion: even if the negation isn’t directly confirmable by sense observation, it still counts as verifiable.
If this is borne in mind, I think that you will find my criticism stronger than you have allowed. Under the revised criterion, every statement that conforms to the rules of syntax can be shown to be satisfy it. The criterion accomplishes nothing.
I still do not consider your argument strong. Let me try and restate it and then explain my objection. First, here is the basic form of the logical positivist criterion of meaning.
You claim that logical positivists to revised their criterion is two ways:
You then assume that P is some non-controversially meaningful statement and that Q is some statement logical positivists would like to reject as meaningless.
It follows from the first revision that if P is meaningful, then ~P is also meaningful. And it follows from the second revision that if P is meaningful, then P v Q is also meaningful.
Your argument then demonstrates that “meaningfulness” is not transmitted from premises to conclusion in a valid argument:
By assumption, Q is meaningless. From the truth of two meaningful premises we have deduced a meaningless conclusion. However, according to the second revision mentioned above, any logical consequence of meaningful statements must itself being meaningful. This contradicts our assumption. The revised criterion of meaning is empty: if there is at least one meaningful statement, then all statements are meaningful.
This argument is right, so far as it goes, but I am unsatisfied. A similar argument would be that, according to this criterion of meaning, contradictions are also meaningful. If P is meaningful, then ~P is also meaningful, and so it follows that so is the conjunction of P and ~P:
Since P & ~P is meaningful and anything follows from a contradiction, it follows that all statements are meaningful. I actually prefer this form of the argument, because it illuminates the underlying error. The logical positivists wanted meaningfulness to be passed from premises to conclusion alongside truth, but, obviously, this means that any attempt to assign “meaningfulness” to contradictory statements will be logically explosive: anything follows.
Why am I unsatisfied with this argument? Because there is a simple fix. We just revise the second second revision:
This prevents the contradiction from ever arising. It also seems to me closer to what logical positivists probably had in mind. Now let us reconsider your argument:
The premise P v Q is not meaningful unless either P or Q have been verified. By assumption, Q cannot be verified, therefore, P must have been verified. But one cannot verify a falsehood and ~P is true by assumption, then P cannot have been verified. Therefore, P v Q is not meaningful after all and the conclusion remains meaningless.
This solution works by tying meaningfulness to logical consequences of statements that have actually been verified and, therefore, truth. Since contradictory premises cannot be verified, even if the premises are two meaningful statements themselves, their logical consequences cannot inherit meaningfulness.
You might consider this argument unfair, since it not the version of the criterion of meaning that you were criticising. But then you use the argument as a way to criticise logical positivism in general rather than a specific version of the criterion of meaning. It is in this sense that I find your argument to lack strength, because a satisfying criticism of logical positivism would have far greater reach: it would apply generally to the idea of a criterion of meaning. Are there such criticisms? I think so. The traditional criticism, associated with Popper, does the trick, i.e. the logical positivist criterion of meaning is unverifiable by sense experience and, therefore, is meaningless by its own standards. Unlike your argument, this criticism cannot be so easily wriggled out of by a simple revision of definition and so, in my view it is a far stronger argument.
To clarify some outstanding issues.
I had assumed that you had made elementary logical errors. I retracted those claims. It seems that we each had different versions of logical positivism in mind, and I shall take your word that your argument faithfully characterises their actual position. I now consider your argument to be sound.
However, I also find the argument unsatisfying. As an explanation of what is wrong with logical positivism, it has very little reach. That is, the criticism applies only to a small subset of positions that could be accurately described as logical positivist, but you present the argument as though it has general applicability.
I have only an elementary understanding of logical, but I think you explained yourself very clearly, and I also got to testout my meagre knowledge of logic, and was educated in the process of reading it.
I meant to say “logical form”.