Unreasonable

Brad DeLong appears to argue here that because pure reason once led him, Brad Delong, to an incorrect conclusion about which direction he was facing, it follows that pure reason can never be a source of knowledge.

(If that’s not his point, then the only alternative reading I can find is that Thomas Nagel is guilty of choosing a poor example to illustrate a point that DeLong would rather ridicule than refute.)

It would be too too easy to make a snarky comment about how we’ve known all along about Brad DeLong’s tenuous relationship with reason. Instead, here, for the record is a list of ten facts, of which I am willing to bet that DeLong is aware of at least 7 — none of them, as far as I can see, accessible to humans via anything but pure reason:

1) The ratio of the circumference of a (euclidean) circle to its radius is greater than 6.28 but less than 6.29.

2) Every natural number can be uniquely factored into primes.

3) Every natural number is the sum of four squares.

4) Zorn’s Lemma is equivalent to the Axiom of Choice (given the other axioms of Zermelo-Frankel set theory).

5) The realization of a normally distributed random variable has probability greater than .95, but less than .96, of lying within two standard deviations of the mean.

6) If p and q are odd prime numbers then the equations x^2 – q = p y and x^2 – p = q y are either both solvable in integers or both unsolvable in integers, unless p and q both leave remainders of 3 when divided by 4, in which case exactly one of them is solvable in integers.

7) If the Peano axioms are consistent, there are true statements about arithmetic that do not follow logically from them (and ditto for any other system of axioms).

8) The Peano axioms are consistent.

9) Every continuous function from the unit disk to itself has a fixed point.

10) The Heisenberg Uncertainty principle follows from the properties of the Fourier transform.

DeLong concludes with this:

And I cannot help but think that only a philosophy professor would believe that our reason gives us direct access to reality. Physicists who encounter quantum mechanics think very differently…

Really? How would you deduce the uncertainty principle from anything but pure reason? How does he think Dirac predicted the positron?

Hat tip to Bob Murphy who called this to my attention. Gene Callahan comments here.

Edited to add: A more pointed Gene Callahan post is here.

Share/Save

49 Responses to “Unreasonable”

1. 1 1 Roger

Statement #1 is accessible with tools other than pure reason. I can draw a circle with a compass, and use a meassuring tape to get the ratio.

Statement #10 is ambiguous. There is a HUP that follows from Fourier principles, but the HUP usually means a physical property of electrons, not a math fact.

Maybe DeLong’s oddest comment is, “Physicists who encounter quantum mechanics think very differently…” He seems to be saying that physicists stop using reason when then learn quantum mechanics. If that were true, then Physics would have ground to a halt in 1925.

2. 2 2 Phil

#1 should be “radius” rather than “diameter”, right?

3. 3 3 Steve Landsburg

Roger: Fair enough re statement #1, but what if I had run these out to a hundred decimal places? Surely DeLong does not own both a measuring tape and a sensory apparatus that will allow him to measure that precisely — but reason still works, out to a hundred decimal places or a million.

Agreed also re #10; I meant the math fact, not the model that applies that fact to physics. But I do continue to maintain that you’d never be aware of this fact except by having reasoned about the mathematical properties of the Fourier transform (or something equivalent).

Agreed also re DeLong’s oddest comment.

And I’ve corrected the Callahan link. Thanks.

4. 4 4 Steve Landsburg

Phil: Good catch. Fixing now.

5. 5 5 Roger

I do not agree that Dirac used pure reason to predict the positron. He was trying to model the proton, and only in a subsequent paper he decided that would not work, and suggested that the equation might describe another particle. Yes, he used reason, but so does every scientist. You could use a better example here.

6. 6 6 Victor Ying

I think DeLong is getting at a defensible point, though. There’s a distinction between knowledge of abstract and artificial concepts and knowledge of physical reality. All 10 of the statements you have listed are mathematical statements that don’t directly tell us anything about reality unless we assume or know from experience or empirical evidence that reality works according to the underlying rules that those statements may be derived from. Knowledge generated purely from reason, and not based on correct empirical knowledge, does not by itself teach us about the workings of the particular reality we live in.

7. 7 7 v

Perhaps you should try to counter something like Devitt’s paper “There is no apriori”, rather than DeLong…

In any case, most if not all of the examples you’ve given use mathematical concepts that have a great deal of practical use. So, the argument goes, the truth of those mathematical statements is not completely disconnected from empirical experience, and it is ultimately derived from this practical utility.

@Roger, you mean electron not proton.

8. 8 8 Sam

Seems odd to me that DeLong refutes circumstantially faulty reason with more a sophisticated reasoning to show that reasoning is not a direct link to objective reality. Confusing the model with logic perhaps?

9. 9 9 Martin-2

DeLong seems to draw a fundamental distinction between knowing something and being very sure. We can be very sure of something through pure reason, but we can’t know it without “direct access to reality”. This could all be made much more precise. DeLong should estimate a threshold of probability past which pure reason can’t take us.

10. 10 10 Ken B

Like Roger I’m not sold on the Dirac/HUP example. After all QM does not treat an electron (or other ‘particle’) as *just* a wave. Nor as *just* a particle. ‘Wave’ is an abstraction. It fits electrons in some ways, as does ‘particle’. This isn’t pure reasons applied to waves, it’s empirically based.

I will also note that the kind of knowledge embodied in 10 is different than the kind embodied in 1-9. It could be wrong.

11. 11 11 Ken B

RE *, the consistency of PA: Didn’t Steve just make a big to-do about rejecting second order logic? But that’s how you *prove* the PA are consistent. Steve’s other argument for consistency is that we ‘know’ the natural numbers intuitively, so have a real model. We’ve been around that tree often enough. I don’t think you can claim we know PA is consistent from pure reason unless you allow second order to count as pure reason.

12. 12 12 Matthew

Who the hell is Brad DeLong?

13. 13 13 Ken B

DeLong’s oddest comment is, “Physicists who encounter quantum mechanics think very differently…” He seems to be saying that physicists stop using reason when then learn quantum mechanics.

He’s saying they rely on observation and experiment not just reason. Most of Steve’s facts are reliant on *just* reason. No part of Quantum Mechanics is founded upon *just* reason.

14. 14 14 Roger

Almost everything in science relies on observation and experiment not just reason. I don’t know why quantum mechanics would be different from anything else.

Dirac was working on an equation for the electron when he found a way to incorporate positive charges. Soon after that the positron was discovered experimentally.

A better example of a quantum mechanics principle developed from nearly pure reason is CPT symmetry. It is almost universally believed as a result of a mathematical argument, but very difficult to test directly.

15. 15 15 Al V.

When it was pointed out to Einstein that the General Theory of Relativity implied that the universe could not be stationary. Assuming a stationary universe, Einstein introduced the cosmological constant. Of course, Hubble showed that the universe was not stationary, and the cosmological constant was unnecessary.

If DeLong had been in Einstein’s shoes, he would have concluded that the conflict between General Relativity and a stationary universe invalidated all of physics, back to Newton and beyond.

16. 16 16 James Knight

Steve, it all depends on what is meant by ‘pure’ reasoning – but if pure is supposed to mean ‘knowledge without needing experience’ then you’re wrong on this one, because everything known requires experience. We have no knowledge without experience – everything known is known from experience (read Hume on this).

All your 10 examples are not “facts accessible to humans via anything but pure reason” – they are “facts accessible to humans with the use of reason from prior experience”

If your definition of ‘pure’ is not ‘knowledge without needing experience’, then I have no problem, but you’d do well to define ‘pure’ in this instance.

17. 17 17 James Knight

Sorry I meant to say – all your 10 examples are not “facts accessible to humans via pure reason” – they are “facts accessible to humans with the use of reason from prior experience”

18. 18 18 Ken B

Roger: We were discussing HUP. I’m just sticking to the topic at hand. Of course science is empirical; that’s DeLong’s point.

Here’s what Delong said (in his “oddest comment”), paraphrased: “Nagel claims we can know the world as it is by pure reason a priori, free of any faults inherent in the way our minds evolved. Scientists who study reality at its most basic and strange disagree.”

19. 19 19 Bob Murphy

Great stuff Steve, but (like some others) I think you oversold your case with the Dirac thing at the end. DeLong is going to say (understandably), “What are you talking about Landsburg?! I can come up with all sorts of internally consistent models of subatomic particles. But the reason we think QED is right, is that it matches experimental observation. You’re setting up a strawman here.”

But, you did NOT set up a strawman with your ten statements. It is quite clear that these are important facts about reality–that’s why we teach them to (some members of) each new generation after all–but they are obviously accessible only through pure reason.

20. 20 20 darf ferrara

Another example that you might cite is Galileo’s demonstration that heavier objects don’t fall faster than lighter objects. This gives another example of a physical fact which can be known by “pure reason”, which is simpler than many of the other examples.

21. 21 21 Ken B

@Bob Murphy: I think everyone is over reading. DeLong thinks Nagel is claiming some infallible access to reality. DeLong says no.
Steve thinks mathematics IS reality, so he’s with Nagel. But if you dispute that identification then you should be with DeLong.

Under 8 Steve is making a circular argument *in this context.* Steve rejects proofs in second order logic. His case for PA being consistent is that N exists and is real. That’s the issue under debate here, so point 8 is circular when used against DeLong.

22. 22 22 Steve Landsburg

Ken B:

Steve rejects proofs in second order logic.

No. I reject the assertion that second order logic is logic. But I take “reaoning” to be something broader than just “formal logical deduction from axioms”.

23. 23 23 Steve Landsburg

James Knight:

All your 10 examples are not “facts accessible to humans via anything but pure reason” – they are “facts accessible to humans with the use of reason from prior experience”

What “prior experience” do you think goes intp my knowledge that every natural number is the sum of four squares?

24. 24 24 Ken B

@darf: That’s a perfect example of a fact that cannot be known a priori.

#9 is interesting in this regard though.

@Steve: Is it possible that there are no unit discs in reality, so that Brouwer’s Fixed point theorem does not have any empirical content?

25. 25 25 Neil

Reason (logic) is used to advance knowledge by projecting the implications of existing knowledge. That is what Dirac’s equation did. Dirac himself did not believe the implication of his own equation (he is famous for saying “the mathematics was smarter than he was”. And the positron (anti-electron) did not become “knowledge” until Carl Anderson observed it.

26. 26 26 Roger

The root of the confusion here is Kant’s Analytic-synthetic distinction. Nagel probably assumes that his readers are familiar with the distinction, but DeLong is not.

27. 27 27 Ken

Steven,

What “prior experience” do you think goes intp my knowledge that every natural number is the sum of four squares?

Your familiarity with the structure of integers, which was informed by your familiarity with simple arithmetic, which was certainly not built on “pure” reasoning, but rather direct experience with counting objects. As an infant, you certainly recognized that two spoonfuls of food was more than one, without formalizing the concepts of “one” and “two”. I’m sure that you don’t think that you have “prior experience” with the integers because you have so internalized them.

As with everyone else’s argument that quantum mechanics is not the result of “pure” reasoning, anything known about the integers isn’t “pure” either. It’s hard to define what “pure” reasoning is as everything we know is based on experience in the actual world. The theorem that all integers are the sum of four squares is solidly built on top of numerical experiments, which are not any more “pure” reasoning than any other type of physical experiment. That the integers are “out there”, it’s clear that they are not the product of pure reason, but of trial and error. What we derive using “pure” reasoning about the integers is based on the axioms derived from the prior experiences we have with integers. If a counter example to any of the theorems were found relating to the integers, no one would, including you, that the integers are busted, but that our reasoning is busted.

This is not to say that nothing can be gotten from reasoning. In fact, I think that much of what we know is built on reason. But it does mean that the idea of “pure” reasoning is very blurry, if not completely meaningless.

28. 28 28 Ken B

“What “prior experience” do you think goes into my knowledge that every natural number is the sum of four squares?”

1.5 billion years of evolution.

29. 29 29 James Knight

Steve >>What “prior experience” do you think goes intp my knowledge that every natural number is the sum of four squares?<<

Your experience of natural numbers, integer numbers, etc – that is the prior experience. It is impossible to know that every natural number is the sum of four squares without prior experience of natural numbers and learned mathematical procedures. Hence, there is no knowledge obtained by pure reason without prior experience.

30. 30 30 James Knight

darf ferrara >>Another example that you might cite is Galileo’s demonstration that heavier objects don’t fall faster than lighter objects. This gives another example of a physical fact which can be known by “pure reason”, <<

Again, this is not pure reasoning – it is reasoning based on experience. To know that heavier objects don’t fall faster than lighter objects one needs prior knowledge of Newtonian laws, which is acquired by experience of the world.

31. 31 31 Steve Landsburg

James Knight:

Your experience of natural numbers, integer numbers, etc

Be careful. I have experiences that teach me about individual numbers like 1, 2, 3 and perhaps even 156,324. That experience is relevant to my knowledge that 1+2=3.

But I have no experience (in the sense in which I think you mean the word) with the *set* of all natural numbers. That is, I have no experience relevant to my knowledge that every natural number is a sum of four squares.

A statement about *particular* natural numbers is of a very different kind than a statement about the collection of *all* natural numbers. So I repeat: What experience do you think I’ve had that might be relevant to a *universally quantified* statement such as “every natural number is a sum of four squares”?

32. 32 32 Ken

Steven,

A statement about *particular* natural numbers is of a very different kind than a statement about the collection of *all* natural numbers.

So? Seeing a pattern for a set of *particular* numbers leads people to see if that pattern is general or merely peculiar to a finite set of numbers.

Can you name a single example of a theorem that characterizes all natural numbers or even just an infinite subset that wasn’t the result of seeing a pattern for *particular* numbers? The open question you like so much is the twin prime conjecture. But that conjecture is only a conjecture because people noticed that no matter how large numbers got, they found a pair of twin primes. The experience of noting this set of *particular* numbers led to asking if there is an infinite number of them.

33. 33 33 Steve Landsburg

Ken: Skewes proved in 1933 that the number of primes less than x is sometimes greater than the function li(x) (the “logarithmic integral” of x) and that the first such example occurs for some x less than the unfathomably large number 10^(10^(10^963))). Whatever it would mean to have “experience” of some such x, Skewes surely didn’t have that experience. Yet he was able, by pure reason, to show that such an x exists. (Indeed, he showed that there are infinitely many such x’s, and indeed that there the functions pi(x) and li(x) cross each other infinitely often, with the first crossing occurring somewhere below 10^(10^(10^963))) — where pi(x) denotes the number of primes less than x.)

On what “experience” could Skewes’s theorem possibly be based?

34. 34 34 Bob Murphy

Steve, if I hadn’t watched X-Men growing up, I bet I couldn’t do algebra.

35. 35 35 Cjohn

If we can’t trust our senses and reason, why is he so certain that he was mistaken about which direction he was actually facing in the first place?

36. 36 36 Ken

Steven,

I think we’re going around in circles. Seriously, do you think that anyone who didn’t have experience computing integrals, or computing pi(x), or computing li(x) could simply pop out a theorem like Skewes did? Gauss thought of the relationship between li(x) and pi(x) a hundred years before Skewes ever heard of either. In other words, a hundreds years (actually thousands) of accumulated mathematical experience and knowledge was passed to Skewes. In addition to what he inherited, he experienced integers, integrals, li(x), pi(x), as all of us have who have encountered them, through experimenting to understand what these things are, then made a conjecture about what he thought to be true.

To advance what you mean, will you please explain what you mean by “pure” reason? Until this is well defined, I’m afraid, we’re just circling each other.

What I claim is that the experience with what is “out there” through basic counting in infancy and toddlerhood, experiencing and experimenting with ever more mathematical objects that are “out there”. These have to be experienced first. Conjecture and reasoning come later. And even then, it’s by experimentation (lets apply theorem A, oops, went no where; now lets apply theorem B, okay, made some progress; and so on and so forth) and trial and error.

You seem to be claiming that that conjecturing, then experimentation is “pure” reasoning, based on NO experience. I think this is wrong, as Skewes himself had all sorts of mathematical experience with numbers, primes, and integrals being a student of Littlewood. If you want to claim that after experiencing LOTS of mathematics, then making a quite reasonable conjecture (that li(x) and pi(x) may intersect) to advance math by quite a small amount compared to he’d all ready experienced, then claim that small amount is “pure” reasoning, fine. I’m not quite sure how you can justify the “pure” part, after spending a life time (34 years in Skewes’ case) experiencing all sorts of mathematical objects.

37. 37 37 Steve Landsburg

Ken: But almost all of the “experience” you’re describing is internal experience — the result of cogitation, not of sensory input. When I was a child, I acquired a lot of “experience” with multiplication, but I’d classify almost all of that experience as taking place in the realm of pure reason.

38. 38 38 Ken B

@Steve re 22:
Then I think you have another problem. You argue that given N, R is not fixed. Keshav Srinivasan showed that if you allow second order, you can get either from the other. So if you accept that bit of second order logic reasoning a major pin of your metaphysics is knocked down.
And as far as the consistency of PA goes, you need no longer appeal to N as a model. That doesn’t knock down a pin but it does mean you can get the consistency of PA without a belief in the reality of N. The obvious consistency of PA though is one of the more convincing of your arguments for the reality of N.

39. 39 39 Martin-2

Ken 36: “do you think that anyone who didn’t have experience computing integrals, or computing pi(x), or computing li(x) could simply pop out a theorem like Skewes did?”

Any theoretical person or any realistic person?

40. 40 40 James Knight

Steve >>A statement about *particular* natural numbers is of a very different kind than a statement about the collection of *all* natural numbers. So I repeat: What experience do you think I’ve had that might be relevant to a *universally quantified* statement such as “every natural number is a sum of four squares”?<<

Indeed, I think what this shows is that mathematics is in a class of its own – that it exists in a reality far beyond, and more complex than, the physical reality with which we interface. I mean, my knowledge that Newton's gravitational laws applies to every planet in the universe is limited relevant to my experience of those planets, because I have only experienced this planet – but mathematics enables me to know it, just like it enables me to know that "every natural number is a sum of four squares". Mathematical laws enable me to know things of which I have limited relative experience.

So while in the physical world we can have no new knowledge without experiencing it, with mathematics we can have knowledge of things like the *set* of all natural numbers without being able to experience them all. For that reason, I think mathematics is in a class of its own.

41. 41 41 Gavin

If “Reason” is understood as the process of properly integrating the resources provided by man’s senses. Then the process of reasoning already presupposes that we have access to reality.

But if reason is to be understood as abstract logic, then reason doesn’t tell us anything about reality. It only tells us if we are thinking correctly about reality.

42. 42 42 Steve Landsburg

Gavin: Rather than make pronouncements that we’re apparently supposed to take as revealed truth, you might try engaging with some of the actual arguments that people have made.

43. 43 43 Gavin

Steve:

I was trying to engage with the arguments by responding to Ken’s call for a definition of Reason. But, based on your comment it seems that I failed. I’ll try again.

Some have interpreted DeLong’s statement to mean,”we cannot obtain knowledge of external reality without ever having prior direct experience with reality.” If he is trying to say that, then he is correct.

You disagree by saying, “What “prior experience” do you think goes into my knowledge that every natural number is the sum of four squares?”

Math is ultimately dependent on direct experience with reality because (1) math is dependent on numbers, (2) numbers are concepts, and (3) concepts are ultimately dependent on perceiving reality.

Reality causes our sensations. A perception is a group of sensations automatically retained and integrated by the brain of a living organism. We simply perceive too many objects to remember every individual thing. We mitigate this problem through conceptualizing our perceptions. A concept is cognitive unit of meaning—a symbol that refers to objects in reality. Humans can form concepts because we can recognize similarities and differences among objects in reality. We summarize these similarities or differences through a process of abstraction.

Example: Green is a concept. Green does not exist by itself just floating somewhere in reality. There is no “green” that one can point to. But we can see a green mango, a green car, and a green turtle. These each have the property green. Green does not exist apart from green objects, but we can abstract the concept of green and talk about the concept of green independently of objects.

Numbers are the same way. There is no number 5 running naked in the wild. We derive the the concept of 5 by observing 5 fingers, or 5 mangos, or 5 turtles. From these experiences we can create the concept of “5 units”. We also observe that 2 units + 3 units = 5 units. Once we understand the number-concepts and how they integrate together, we can reapply mathematical concepts and rules back on the concepts themselves. Through this process we can gain knowledge without having further experiences. As Wittgenstein said, “From the given, I can construct what is not given.”

The “prior experiences” that lead to the knowledge that every natural number is the sum of four squares is not some experience about natural numbers and squares running around in the wild. The prior experiences were the perceptions in your youth that led to the creation of concepts of numbers. The prior experiences also include hearing your teachers tell you how to integrate these number-concepts correctly.

This process of integrating concepts is simply called reason. To put it more clearly, Reason is the ability to integrate concepts as derived from the senses. Our sensations are sensations about reality. Since reason already presupposes access to reality, it is contradictory to say (as Brad DeLong does) that reason can’t give us access to reality.

44. 44 44 Jamie Whyte

deLong’s argument against Nagel is hopeless. Set it aside.

The interesting question is whether our knowledge of logic shows that “the materialist, neo-Darwinian conception of nature is almost certainly false” (the subtitle of Nagel’s book). I have not read the book; I know only the quoted passage from Gene Callahan’s (sympathetic) post:

“I see that the contradictory beliefs cannot all be true, and I see it simply because it is the case. I grasp it directly.”

How Nagel gets from here to his sub-title, I do not know (because I have not read the book, sorry). But this passage makes me think that he is relying on a linguistic trick. The idea that some kind of non-physical interaction (between logical facts and minds) is occurring here arises from his use of the word “because” (which suggests causation) and “grasps”, which also suggests some kind of contact.

Let me describe the situation in a way that gives rise to no such suggestion of non-physical causation. I am sitting behind a veil and, on the other side, someone is tossing a coin. I am asked to say whether each toss lands heads. If my success rate, over the long run, is 0.5, we will feel no need to posit any mechanism of apprehension. I do not apprehend. If I get a long run success rate well above 0.5, then we may start hypothesizing about extra-sensory perception and so on. It is precisely because people cannot do better than 0.5 in these circumstances that most of us do not believe in ESP.

Now suppose that the probability of the outcome I am being asked about from behind the veil is 1, perhaps because it is logically certain. Suppose, for example, that I am told that a number of quadrupeds will be brought to the other side of the veil and I must say which of them have four legs. Even if I have a 100% success rate, no one will believe that some kind of weird non-physical apprehension is going on, that I am getting the answers right because I am “grasping” the features of the quadrupeds behind the veil.

You may wonder how come I understand that all quadrupeds have for legs or, more generally, you may wonder how I reason logically, but the answer does not require me to be “seeing” what is behind the veil. Can a Darwinian explain how come human minds (not all of them always!) are apt to reason in accordance with modus ponens, modus tollens, etc.? I think it can.

Once you have a physical or Dawinian explanation of our logical abilities, combined with the necessity of logical and mathematical truths, you are going to find it difficult to create a case for non-physical apprehension or non-physical causation from the fact that we know logical facts.

45. 45 45 Harold

Our brains develop as we grow. They do this by interaction with the world. This interaction strengthens some connections and links within the physiology of our brains. Without this experience we would not be able to use these connections to reason about mathematical abstractions. This is the experience you need before you can arrrive at conclusions by pure reasoning. No person can therefore arrive at these conclusions without experience, but that is not the same as saying that these conclusions can not (in principle) be arrived at without experience. The experience is needed to build the machine, as it were.

It is interesting to speculate about how the nature of the interactions we have whilst building our minds limits the type of reasoning we can do. I am not sure if Steve is saying that there are no limits, as once we have acheived reason it can theoretically take us anywhere.

As for Nagel, he is talking crap. “conscious minds and their access to the evident truths of ethics… ” ??? Anyone who uses that line is to be treated with a great deal of suspicion.

46. 46 46 Ken B

@Jamie Whyte: We need to see you here more often!

I am assuming you are the author of the entertaining Crimes Against Logic. I have used it as a stocking stuffer!

47. 47 47 Jamie Whyte

@Ken B

Good on you! as we say in New Zealand. And just remember, its Xmas time again!

48. 48 48 Keshav Srinivasan