Sunday, July 11, 2010

Guttenberg and Synthetic a Priori

Mattheus von Guttenberg has probably his clearest articulation yet of a priorism and Austrianism. He starts with a review of Kant's four part explanation of knowledge - the intersection between synthetic/analytic and a priori/a posteriori. Guttenberg refreshingly does not just fall back on the claim that analytic a posteriori is some kind of impossibility - he gives an example and moves on. Most of the discussion of course focuses on the synthetic a priori and Mises.

I want to briefly play Friedman to Guttenberg's Mises. Mises's logical deductions are fine. Of course no economist objects to synthetic a priorism. But does it follow from the soundness of true synthetic a priori knowledge that empiricism is useless for the practice of a priorism? I don't think it does. Guttenberg writes of Mises:

"All his conclusions arrive from logical deductive processes, and are therefore logically sound. Empirical evidence is useless to challenge these views, they must be argued via reason if one finds them faulty."

I would have rewritten the first sentence as:

"All his conclusions arrive from an attempt at logical deductive processes from asserted axioms, and therefore are logically sound if the attempt is successful"

I don't think Guttenberg would object to the addition of this caveat. To follow in the Kantian spirit, my revision of that sentence is an analytic a priori statement (or is it? one could critique my "classical foundationalism" in this claim, as it was critiqued in this post). My revision of the first sentence immediately reveals the problem of the second sentence. Since Mises's soundness is contingent on his argument, the usefulness of empiricism is also contingent on his argument. True synthetic a priori statements cannot be challenged by empirical evidence because that would imply a logical contradiction. The point is, we're not dealing with "true synthetic a priori statements" here - we're dealing with a whole book full of statements that may or may not be "true synthetic a priori statements" written by an economist in 1949.

The General Theory of Employment, Interest, and Money is another synthetic a priori work of economics. It has a few pieces of data in it, but they're all used to illustrate points rather than derive knowledge. Clearly, though, Keynes and Mises are at odds on some points. Guttenberg would suggest that we should arbitrate between Keynes and Mises on purely logical grounds. Fine. I say Mises neglects an important axiom (or else fails to derive it from the single axiom he does use). Mises notes that "man prefers sooner rather than later", but he neglects "man prepares for unforeseen contingencies" - i.e., liquidity preference. One could make some kind of argument about how "man prepares for unforeseen contingencies" is an axiom. Or one could make an argument that it can be derived from Mises's axiom, that "man acts". I think the second is probably the better option. The point is, a lot of the differences between Austrianism and Keynesianism come back to Keynes's use and the Austrian's neglect of liquidity preference. Guttenberg wants to arbitrate this with logic, and I consider the Austrians to be on the losing end of the logic. They've neglected an essential synthetic a priori point and it screws up the rest of their system. Guttenberg is obviously going to deny there is a logical problem at all. So where does this get us? Nowhere. What's the solution?

This is why empiricism is so important for any science. Human minds are not perfect, and we make mistakes. Obviously either I am wrong about economics or Guttenberg is. Or, perhaps more likely, both of us are. But what we do know (barring, of course, those critiques of classical foundationalism) is that even if Guttenberg and I can't agree on what constitutes true synthetic a priori knowledge, we both can agree that true synthetic a priori knowledge and true synthetic a posteriori knowledge can not contradict each other. We can also agree that false claims that claim to be synthetic a priori knowledge and true synthetic a posteriori knowledge may contradict each other. It seems to me, then, that since the whole question is whether Mises gives us real synthetic a priori knowledge or not, one potential way of determining the question is testing it against synthetic a posteriori evidence. Since neither Guttenberg nor I will give ground on the synythetic a priori front, and since we have a sense of the non-contradictory nature of different types of knowledge, empiricism should be useful for arbitration if you accept Kantian epistemology.

Of course, practical attempts at synthetic a posteriori knowledge (like a linear regression run in 2010) may be as faulty as practical attempts at synthetic a priori knowledge (like bounded collections of statements written in 1936 or 1949). We always run the risk of being wrong in any of these endeavors. The question is, why does Guttenberg privelege one endeavor over another (a priori over a posteriori), when the problem is so obviously that we have two synthetic a priori theories that contradict each other but that neither of us will concede logical fault on?

Another way of making my point is to say that OK, true synthetic a priori statements and true synthetic a posteriori statements cannot contradict. So if real-world attempts at synthetic a priori statements and real-world attempts at synthetic a posteriori statements don't contradict we have good reason to think both may be true. But if they do contradict then at least one of them, and perhaps both, are faulty. My question to Guttenberg is why, in the case of such a contradiction, is he so quick to assume that it is the real-world attempt at the synthetic a priori that is true? How does he arbitrate between those two? Kant gives us no way to arbitrate - so instead I think we're forced to be pragmatic and corroborate. Guttenberg only feels comfortable with his position because he doesn't see both the synthetic a priori and synthetic a posteriori claims as being imperfect, potentially faulty, real-world attempts at knowledge. He assumes that one is and one isn't. He assumes that the synthetic a posteriori breaks on the synthetic a priori. I'm saying that that assumption has no basis.

I could rewrite Guttenberg's whole post and replace references to Mises with references to Keynes. Would that convince him? Of course it wouldn't. Appeals to Kant don't settle the question because we all more or less agree with Kant. Kant was not an Austrian (he was a Prussian, actually!).

17 comments:

  1. I must be moving up in this world if you're referring to me by my surname now. I don't know which I like more.

    Let me preface the responses by saying I think this is a great post, and great posts lead to great thinking. So in the spirit of thinking deeply,

    RE "I would have rewritten the first sentence as:......"

    Sure, that sounds fine. There's literally no difference. Your inclusion of specific phrases (asserted axioms) takes away the grandness of the statement, but there is nothing incorrect about the initial axiom and so your sentence and my sentence are both correct.

    RE "The point is, we're not dealing with "true synthetic a priori statements" here - we're dealing with a whole book full of statements that may or may not be "true synthetic a priori statements" written by an economist in 1949."

    I respectfully disagree. To my knowledge, no economist has ever analyzed economic phenomena by the lens of logic in this sense as Mises has done. Keynes made claims and perhaps skillfully supported them, but with Mises it is clear where he gets them from, and more importantly how he gets to conclusions.

    The first third of Human Action is roughly an introductory course on epistemology. There is no economics in the first volume.

    RE "Mises notes that "man prefers sooner rather than later", but he neglects "man prepares for unforeseen contingencies" - i.e., liquidity preference."

    He does no such thing. You habitually bring this up to strike a distinction without a difference. Mises does not use the words 'liquidity preference', but his entire theory of interest is time-based. The notion of originary interest cannot be understood without reference to "sooner" rather than "later." He almost explicitly proves the case for liquidity preference, probably in a light just as strong as Keynes does. We just don't attach an enormous significance to it considering the vastness of his work.

    RE "...empiricism should be useful for arbitration if you accept Kantian epistemology."

    Yes and no. I agreed with everything you said on how different modalities of knowledge must still be consistent. That being said, in the realm of economics, (as I have stated elsewhere) the strength of empiricism to deliver true claims is very weak.

    Consider all the many schools of economics. Of all of them, how many are empiricism-based? Almost all? And what kind of variance is there between the schools who all claim to study the same topic?

    When the topic is simple, ie nonchanging, like analyzing chemical compositions, it is useful to approach from empirical methods because the amount of variables, and the strength of those variables, are themselves weak.

    Another reason I disagree with empiricism in economics is from what we've learned in regards to quantum mechanics. The simple act of observation tends to change conclusions. If you want to study how man acts on his own, for instance, it cannot be done empirically because that study will change his behavior. Although, I do agree that in principle, true synthetic a posteriori must agree with true synthetic a priori - the question is one of practice. Which is more likely to lead one to many differing conclusions?

    ReplyDelete
  2. RE "when the problem is so obviously that we have two synthetic a priori theories that contradict each other but that neither of us will concede logical fault on?"

    Really? Where does Keynes say or even imply that his statements about economics derive from objectively true synthetic a priori axioms? I've never heard that one before, Daniel. If it were the case that Keynes wrote his book on synthetic a priori grounds (as opposed to his earlier books which he explicitly repudiates in GT), then we must do as logicians do and attempt to find flaws in either's work.

    RE "He assumes that one is and one isn't."

    Careful now. I never actually did make this assumption, and I would not agree with it if someone were to assume this. You're quick to assume I take sides on these issues, but oftentimes I take no side.

    If I were to write my article on "The Synthetic A Priori and What It Means for Electrical Engineering" I would be quite in the wrong to be praising aprioristic methods to answering tough questions related to a simple, empirical science. The decision - to follow apriorism or empiricism - is a question of difficulty of the science itself. In a simple science such as biology (where subjects do not act differently under observation), we can isolate variables and predict with near certainty how things will behave. Acting man is multitudes more complex.

    Really now, I feel very childish saying all this because Mises has a chapter in Human Action dealing with this EXACT question. Chapter II deals with "Epistemological Problems in Economics" http://mises.org/pdf//humanaction/pdf/ha_02.pdf

    The only thing I didn't mention in my article, which is of crucial importance to Misesian praxeology is methodological individualism. There was no way I tie that in with Kant, so I left it aside. That might answer more questions.

    ReplyDelete
  3. "Praxeology is a priori. All its theorems are products of deductive reasoning that starts from the category of action. The questions whether the judgments of praxeology are to be called analytic or synthetic and whether or not its procedure is to be qualified as "merely" tautological are of verbal interest only.

    What praxeology asserts with regard to human action in general is strictly valid without any exception for every action. There is action and there is the absence of action, but there is nothing in between. Every action is an attempt to exchange one state of affairs for another, and everything that praxeology affirms with regard to exchange refers strictly to it. In dealing with every action we encounter the fundamental concepts end and means, success or failure, profit or loss, costs. An exchange can be either direct or indirect, i.e., effected through the interposition of an intermediary stage. Whether a definite action was indirect exchange has to be determined by experience. But if it was indirect exchange, then all that praxeology says about indirect exchange in general strictly applies to it.

    Every theorem of praxeology is deduced by logical reasoning from the category of action. It partakes of the apodictic certainty provided by logical reasoning that starts from an a priori category.

    Into the chain of praxeological reasoning the praxeologist introduces certain assumptions concerning the conditions of the environment in which an action takes place. Then he tries to find out how these special conditions affect the result to which his reasoning must lead. The question whether or not the real conditions of the external world correspond to these assumptions is to be answered by experience. But if the answer is in the affirmative, all the conclusions drawn by logically correct praxeological reasoning strictly describe what is going on in reality."

    ReplyDelete
  4. "The a priori categories are the mental equipment by dint of which man is able to think and to experience and thus to acquire knowledge. Their truth or validity cannot be proved or refuted as can those of a posteriori propositions, because they are precisely the instrument that enables us to distinguish what is true or valid from what is not."

    ReplyDelete
  5. The Ultimate Foundation of Economic Science... I need to add that to my book list.

    Thanks, Anon.

    ReplyDelete
  6. First, Keynes doesn't have as formal a system of a priorism as Mises because that wasn't his project in the way that it was Mises's project. I'll be happy to concede that. It makes the comparison a little rougher, but I don't see the major problem with it.

    Although, I do agree that in principle, true synthetic a posteriori must agree with true synthetic a priori - the question is one of practice. Which is more likely to lead one to many differing conclusions?

    That's exactly my point! I suppose we just answer the question differently. I think both are likely to lead to many differing conclusions, ultimately. Which is "more likely" probably depends on the question, and I see no reason to believe that a priorism has a consistent advantage. Empiricism has a major advantage - it can be done multiple times.

    See, I think you turn the issue on its head. If economics were a simple, clean science like math you could use pure a priori. Some mathematicians now use quasi-empirical methods for tougher problems where the issues at hand aren't as clean-cut, but obviously most of it is a priori knowledge. When you get into more complicated phenomena that the human mind simply can't capture, more empiricism is introduced - like in chemistry or physics. The science of truly complex systems: biology, and the study of human society - have to rely even more heavily on empiricism because the processes are just too complex to faithfully render with a priorism.

    As Alfred Marshall once said - nature's action is complex, and nothing is gained by pretending that it is simple, and reducing it to a series of elementary propositions.

    ReplyDelete
  7. He does no such thing. You habitually bring this up to strike a distinction without a difference. Mises does not use the words 'liquidity preference', but his entire theory of interest is time-based.

    What does being time-based have to do with liquidity preference? Who doesn't have a theory of interest that is time-based?

    Consider all the many schools of economics. Of all of them, how many are empiricism-based? Almost all? And what kind of variance is there between the schools who all claim to study the same topic?

    Hmmmm... most schools of thought aren't differentiated on empirical issues. They're differentiated on theoretical issues. Now, perhaps the difference is derived from the fact that their non-empirical theorizing isn't solid (which isn't surprising - because the human mind has a lot more wiggle room in its theoretical explorations than in its empirical explorations), but it's not derived from anything having to do with empiricism.

    This introduces another problematic issue for this discussion - that pure theory is not exactly the same as a priorism, although they have a lot of similarities. Theory justifies itself as being a starting point of sorts for empirical investigations, and as such it is entirely outside of anything (well anything that I'm aware of) that Kant got into.

    ReplyDelete
  8. See, I think you turn the issue on its head. If economics were a simple, clean science like math you could use pure a priori. Some mathematicians now use quasi-empirical methods for tougher problems where the issues at hand aren't as clean-cut, but obviously most of it is a priori knowledge.

    Right. And of all the sciences, mathematics is the one that Kant often went back to as being this sort of pure science.

    I'm not familiar with the thought of Mises, but is he basically trying to establish economics as a logical outworking from basic principles like "man acts"? It almost seems like his project is economics-as-metaphysics, if this is how he argues for its basis as a science.

    I'm also not sure about some of your definitions, Mattheus. For instance, that synthetic propositions are "claims that deal specifically with the real world"... the definition more relevant to Kant's discussion of the synthetic a priori should probably have something to do with adding new knowledge of the subject that is not contained in the more basic concepts that are the constituent parts of such claims. So, "2 + 2 = 4" is synthetic, and I think Kant would say that your proposition about triangles is too (rather than analytic, as you have it). Mathematics is for him an example of a science based on synthetic a priori truth, isn't it? And his main project in the Critique... to establish the possibility of metaphysics as science... is based on synthetic a priori truth that does a much more basic job than what you're saying Mises pursues, isn't it? I mean, Kant is concerned with arguing that our perception of appearances requires space and time as subjective conditions, right?

    The question also needs to be addressed sooner or later how one deals with those who disagree with you. You seem, like Kant, to be arguing that a lot of this is just clearly true (not that this is the sum of Kant's argument, of course... I'm not sure how exhaustive it is of Mises' argument). That's fine... I find a good deal of it compelling as well... but the question of whether there is a synthetic a priori at all is still open for plenty of people (or rather, closed because rejected). Much of what Kant has asserted about mathematics has been disputed since non-Euclidean geometry. And his attempt to re-construct morality and metaphysics after his demolition job of previous "dogmatic" approaches has been consistently met with a good bit of skepticism since the time he proposed it. I guess I don't see how Mises wouldn't fall into something like the same fate. And as Daniel brings up regarding mathematics, the move of everything since Kant has been away from laws, away from necessary logical grounds, and more towards various empirical and experimental frontiers. I found your post quite interesting, and really it impressed on me the extent to which these sorts of economic theories remain very much an Enlightenment project. My sense, however, is that scientific inquiry and method has progressed quite a bit since that time, and it seems odd for someone in the 20th century to have anchored their system to something like this, rather than more recent work in the sciences.

    ReplyDelete
  9. ...oh, I forgot to add, doesn't Kant (contra Block) deny the possibility of the analytic a posteriori? It also strikes me as the understatement of the year that Block says so nonchalantly, "Although there is some question about this in philosophical circles" about his one-for-one translation example for a supposed analytic a posteriori. I'm not even sure how one would talk about analyticity alongside translation... one could speak of definition, perhaps... that's where the "all bachelors are unmarried" example gains some traction... but translation seems a much more complicated issue.

    ReplyDelete
  10. “Empiricism has a major advantage - it can be done multiple times.”

    While logical processes can only be thought through once or twice, max?

    “When you get into more complicated phenomena that the human mind simply can't capture, more empiricism is introduced - like in chemistry or physics. The science of truly complex systems: biology, and the study of human society - have to rely even more heavily on empiricism because the processes are just too complex to faithfully render with a priorism.”

    If the human mind can’t capture it, then we shouldn’t be bothering with elucidating the truth of its claims. After all, in order to process empiricism, you NEED a theory behind it otherwise you’re just looking at a lot of data with no way to interpret it. And if we can’t understand it at all, then both modalities are worthless and we’ll never come to terms with it.

    The better way to understand complex systems is not by directly observing millions of variables interacting in unknown ways, but to understand each component logically. Mises creates his evenly rotating economy as a thought creation – to understand the difference between profit and interest – because he knows that the economy is far too complex to understand at face value.

    You can’t dismiss apriorism on any grounds, because then you have no theory to resort to for interpretation.

    As Alfred Marshall once said - nature's action is complex, and nothing is gained by pretending that it is simple, and reducing it to a series of elementary propositions.

    Nobody is suggesting simplicity. In fact, much of the work done by empirical circles argues for simplicity. It is because economics is complex that it requires careful analysis of each individual component. To learn how forests grow, we study the biology of trees.

    What does being time-based have to do with liquidity preference? Who doesn't have a theory of interest that is time-based?

    Frank Knight. His theory is productivity-based. Time has everything to do with liquidity preference because LP is only important insofar as we are uncertain about the future.

    …pure theory is not exactly the same as a priorism…

    If you are doing it right, it is. A priori truths must be known before any evidence is gathered. Don’t confuse “hypothesis” with pure theory – a hypothesis is a guess to be tested against evidence; it necessarily comes after data collection, and as such is inapplicable to the study of economics. Praxeologists do not hypothesize – they conclude logically.

    ReplyDelete
  11. It almost seems like his project is economics-as-metaphysics, if this is how he argues for its basis as a science.

    That's exactly right. Metaphysics is revealed by epistemology - and that's what we're discussing.

    For instance, that synthetic propositions are "claims that deal specifically with the real world"... the definition more relevant to Kant's discussion of the synthetic a priori should probably have something to do with adding new knowledge of the subject that is not contained in the more basic concepts that are the constituent parts of such claims. So, "2 + 2 = 4" is synthetic, and I think Kant would say that your proposition about triangles is too (rather than analytic, as you have it).

    I didn't give the specific Kantian definition because it would cause too much unnecessary confusion. My definition suffices. As far as you contesting my examples, I believe I am correct. 2+2=4 is analytic because the conclusion didn't give us anything new that the premises didn't imply. We define 4 in terms of 2, so we are speaking of numerical tautologies.

    Mathematics is for him an example of a science based on synthetic a priori truth, isn't it? And his main project in the Critique... to establish the possibility of metaphysics as science... is based on synthetic a priori truth that does a much more basic job than what you're saying Mises pursues, isn't it?

    Kant is incorrect on that count. Mathematics is purely analytic because the premises imply the conclusions. The conclusions are true by definition, and so they give us nothing new. I wrote about this in my article Thoughts on Empiricism and Logic on ET.net.

    I'm not familiar with Critique. I have yet to sit down with it.

    the question of whether there is a synthetic a priori at all is still open for plenty of people

    Not at all. There either is or there isn't. I have explained in detail why synthetic a priori conclusions are objectively true. Denial is contradictory.

    My sense, however, is that scientific inquiry and method has progressed quite a bit since that time, and it seems odd for someone in the 20th century to have anchored their system to something like this, rather than more recent work in the sciences.

    I suppose the failure of the "recent sciences" to understand and predict economic events propels people to look for truth, then. I believe the underlying methodology is flawed under Keynesian/Classical/Neo-Classical/Monetarist views because they are not rigorous in the application of logic.

    ...oh, I forgot to add, doesn't Kant (contra Block) deny the possibility of the analytic a posteriori?

    Yes, he does. But as I have shown you, it is quite possible to arrive at a tautological conclusion by a posteriori methods.

    ReplyDelete
  12. So is "praxeology" just a fancy word for "ethics"?

    ReplyDelete
  13. Not at all. Praxeology is a method of discovering truth.

    ReplyDelete
  14. I can't speak to philosophy, but in mathematics both the real number line (inclusive of 2 and 4 and the distance between them) is considered axiomatic.

    Mattheus I'm curious about your conception of what constitutes a "complex system". I may post on that tomorrow morning. But what exactly does a "complex system" mean to you?

    ReplyDelete
  15. The phrase "complex system" isn't any more specific than "large tree" or "deep water." It's my opinion on a topic. I happen to think economics, at face value - that is, when confronted with the bewildering mass of data - is too complicated to discern understanding from it outright.

    Now clearly - I don't suggest that empiricists watch planets orbiting and suddenly BAM, they understand universal gravitation - they solve the problem of computing individual planetary behavior and put it together - but the process by which you and I arrive at conclusions regarding economics are on opposite ends. I consider your process to be useless just as any astronomer would consider the above process to be.

    ReplyDelete
  16. 2+2=4 is analytic because the conclusion didn't give us anything new that the premises didn't imply. We define 4 in terms of 2, so we are speaking of numerical tautologies.

    Yes, we are defining 4 in terms of 2 here. But the reason (as I understand Kant's point) it is synthetic is that you can't get 4 simply by thinking 2. Something is analytic when one can arrive at further conclusions simply by analyzing a concept. To take the common example (although I'm inclined to agree with Quine here that there really is no significant analytic/synthetic divide in the first place) the string-of-symbols "bachelor" is (for English speakers) defined in terms of "unmarried man". This is a tautology of sorts... it's definitional. You can get to "unmarried man" by thinking "bachelor" and vice versa, as long as you are a fluent English speaker. You can't do the same with 2 and 4. Try as hard as you might scrutinizing "2", you aren't going to come up with "4" as any sort of necessary consequence, deeper essence, or definitional convention. That's why you need a "+ 2" to complete the formula... and this is why it's synthetic, because 4 isn't simply a further elucidation of 2, but is rather the conclusion of an arithmetic formula.

    I think when you say the conclusion didn't give us anything that the premise didn't "imply", you're stretching the usage of "imply" a bit to mean something like what goes on in an analytic proposition. But does 2 really imply 4 in this way? Does the number 2 also hold within itself all other even numbers by implication? I don't think so. To reach these sorts of conclusions, you need to add something to the original concept, putting this in the category of the synthetic rather than the analytic (although, to repeat, I'm not entirely convinced that there is much of, if any, analytic/synthetic distinction in the first place. All that I've said above is an attempt to speak for Kant and what I take to be traditional understandings of this distinction).

    ReplyDelete
  17. Me: "You can get to "unmarried man" by thinking "bachelor" and vice versa, as long as you are a fluent English speaker."

    I probably shouldn't have said "and vice versa". While it may be true in this case, I don't want to confuse matters by making it sound like this needs to be reversible for it to be analytic. I don't think it does, does it? The relevant question is whether we can get to the concept "4" simply by thinking the concept "2". Getting from 4 back to 2 simply by thinking 4 seems like it's probably a different question, or at least a question that should be able to be answered separately from the first one.

    ReplyDelete

All anonymous comments will be deleted. Consistent pseudonyms are fine.