I don’t think Deutsch means that mathematical proofs are all inductive. I think he means that proofs are constructed and checked on physical computing devices like brains or GPGPUs; and that because of that mathematical knowledge is not in a different ontological category than empirical knowledge.
I feel quite confident saying that mathematics will never undergo paradigm shifts, to use the terminology of Kuhn.
It believe it already has. Consider the Weierstrass revolution. Before Weierstrass, it was commonly accepted that while continuous functions may lack a derivative at a set of discrete points, it still had to have a derivative somewhere. Then Weierstrass developed a counterexample, which I think satisfies the Kuhnian “anomaly that cannot be explained within the current paradigm.”
Another quick example: during the pre-War period, most differential geometry was concerned with embedded submanifolds in Euclidean space. However, this formulation made it difficult to describe or classify surfaces—I seem to believe but don’t have time to verify that even deciding whether two sets of algebraic equations determine isomorphic varieties is NP-hard. Hence, in the post-War period, intrinsic properties and descriptions.
EDIT: I was wrong, or at least imprecise. Isomorphism of varieties can be decided with Grobner bases, the reduction of which is still doubly-exponential in time, as far as I can tell. Complexity classes aren’t in my domain; I shouldn’t have said anything about them without looking it up. :(
Reading the wiki page, it looks like Weierstrass corrected an error in the definition or understanding of limits. But mathematicians did not abandon the concept of limit the way physicists abandoned the concept of epicycle, so I’m not sure that qualifies as a paradigm shift. But I’m not mathematician, so my understanding may be seriously incomplete.
I can’t even address your other example due to my failure of mathematical understanding.
Reading the wiki page, it looks like Weierstrass corrected an error in the definition or understanding of limits.
Hindsight bias. The old limit definition was not widely considered either incorrect or incomplete.
But mathematicians did not abandon the concept of limit the way physicists abandoned the concept of epicycle, so I’m not sure that qualifies as a paradigm shift.
They abandoned reasoning about limits informally, which was de rigeur beforehand. For examples of this, see Weierstrass’ counterexample to the Dirichlet principle. Prior to Weierstrass, some people believed that the Dirichlet principle was true because approximate solutions exist in all natural examples, and therefore the limit of approximate solutions will be a true solution.
Hindsight bias. The old limit definition was not widely considered either incorrect or incomplete.
Not true. The “old limit definition” was non-existent beyond the intuitive notion of limit, and people were fully aware that this was not a satisfactory situation.
We need to clarify what time period we’re talking about. I’m not aware of anyone in the generation of Newton/Leibniz and the second generation (e.g., Daniel Bernoulli and Euler) who felt that way, but it’s not as if I’ve read everything these people ever wrote.
The earliest criticism I’m aware of is Berkeley in 1734, but he wasn’t a mathematician. As for mathematicians, the earliest I’m aware of is Lagrange in 1797.
That’s pretty clear, thanks. Obviously, experts aren’t likely to think there is a basic error before it has been identified, but I’m not in position to have a reliable opinion on whether I’m suffering from hindsight bias.
Still, what fundamental object did mathematics abandon after Weierstrass’ counter-example? How is this different from the changes to the definition of set provoked by Russell’s paradox?
I don’t recall where it is said that such an object is necessary for a Kuhnian revolution to have occurred. There was a crisis, in the Kuhnian sense, when the old understanding of limit (perhaps labeling it as limit1 will be clearer) could not explain the existence of e.g., continuous functions without derivatives anywhere, or counterexamples to the Dirichlet principle. Then Weierstrass developed limit2 with deltas and epsilons. Limit1 was then abandoned in favor of limit2.
Wikipedia gives the acceptance of non-Euclidean geometry as a “classical case” of a paradigm shift. I suspect that there were several other paradigm shifts involved from Euclid’s math to our math: for instance, coordinate geometry, or the use of number theory applied to abstract quantities as opposed to lengths of line segments.
Reading Wikipedia, it looks like a naive definition of a set turns out to be internally inconsistent. Does that mean the concept of set was abandoned by mathematicians the way epicyles have been abandoned by physicists? That’s not my sense, so I hesitate to say redefining set in a more coherent way is a paradigm shift. But I’m no mathematician.
Its a matter of degree rather than an absolute line. However, I would say a time when even the very highest experts in a field believed something of great importance to their field with quite high confidence, and then turned out to wrong, probably counts.
This looks very like trying to define away something that sure felt like a paradigm shift to the people in the field. Remember that “paradigm” is a belief held by people, not a property inherent in the universe.
Perhaps this is a limitation of my understanding of Kuhn, in that I’m misusing his terminology. I am unaware of mathematics abandoning fundamental objects as inherently misguided the way physics abandoned epicycles or impetus. I expect physics will have similar abandonments in the future, but I expect mathematics never will. The difference is a property of the difference between mathematics and empirical facts. This comment makes the argument I’m trying to assert in slightly different form.
Isn’t that exactly what happened? The phrase “set of all sets that do not contain themselves” isn’t really expressible in Zermelo-Fraenkel set theory, since that has a more limited selection of ways to construct new sets and “the set of everything that satisfies property X” is not one of them.
As I understand it, a paradigm shift would include the abandonment of a concept. That is, the concept cannot be coherently expressed using the new terminology. For example, there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus. I think Kuhn would say that these examples are evidence that empirical science is socially mediated.
I’m not aware of any formerly prominent mathematical concepts that can’t even be articulated with modern concepts. Because mathematics is non-empirical and therefore non-social, I would be surprised if they existed.
A totally trivial nit pick, I admit, but there’s no such thing as the Aristotelian theory of impetus. The theory of impetus was an anti-Aristotelian theory developed in the middle ages. Aristotle has no real dynamical theory.
there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus
There are perfectly fine ways to express those things. Epicycles might even be useful in some cases, since they can be used as a simple approximation of what’s going on.
The reason people don’t use epicycles any more isn’t because they’re unthinkable, in the really strong “science is totally culture-dependent” sense. It’s because using them was dependent on whether we thought they reflected the structure of the universe, and now we don’t. Ptolemy’s claim behind using epicycles was that circles were awesome, so it was likely that the universe ran on circles. This is a fact that could be tested by looking at the complexity of describing the universe with circles vs. ellipses.
So this paradigm shift stuff doesn’t look very unique to me. It just looks like the refutation of an idea that happened to be central to using a model. Then you might say that math can have no paradigm shifts because it constructs no models of the world. But this isn’t quite true—there are models of the mathematical world that mathematicians construct that occasionally get shaken up.
My point was that trying to express epicycles in the new terminology is not possible. That is, modern physicists say, “Epicycles don’t exist.”
Obviously, it is possible to use sociological terminology to describe epicycles. You yourself said that they were useful at times. But that’s not the language of physics.
Since you mentioned it, I would endorse “Science is substantially culturally dependent”, NOT “Science is totally culturally dependent.” So culturally dependent that there is not reason to expect correspondence between any model and reality. Better science makes better predictions, but it’s not clear what a “better” model would be if there’s no correspondence with reality.
I brought all this up not to advocate for the cultural dependence of science. Rather, I think it would be surprising for a discipline independent of empirical facts to have paradigm shifts. Thus, the absence of paradigm shifts is a reason to think that mathematics is independent of empirical facts.
If you don’t think science is substantially culturally dependent, then there’s no reason my argument should persuade you that mathematics is independent of empirical facts.
And so for the curves in question, the Fourier expansion would have only a finite number of terms.
The point being that, in contrast to what was being asserted, Ptolemy’s concept is subsumed within the modern one; the modern language is more general, capable of expressing not only Ptolemy’s thoughts, but also a heck of a lot more. In effect, modern mathematical physics uses epicycles even more than Ptolemy ever dreamed.
My point was that trying to express epicycles in the new terminology is not possible.
But it is! You simply specify the position as a function of time and you’ve done it! The reason why that seems so strange isn’t because modern physics has erased our ability to add circles together, it’s because we no longer have epicycles as a fundamental object in our model of the world.
So if you want the copernican revolution to be a paradigm shift, the idea needs to be extended a bit. I think the best way is to redefine paradigm shift as a change in the language that we describe the world in. If we used to model planets in terms of epicycles, and now we model them in terms of ellipses, that’s a change of language, even though ellipses can be expressed as sums of epicycles, and vice versa.
In fact, in every case of inexpressibility that we know of, it’s been because one of the ways of thinking about the world didn’t give correct predictions. We have yet to find two ways of thinking about the world that let you get different experimental results if you plan the experiment two different ways. In these cases, the paradigm shift included the falsification of a key claim.
Rather, I think it would be surprising for a discipline independent of empirical facts to have paradigm shifts
I don’t think it’s necessarily true (for example, you can imagine an abstract game having a revolution in how people thought about what it was doing), but it seems reasonable for math, depending on how you define “math.” I think people are just giving you a hard time because you’re trying to make this general definitional argument (generally not worth the effort) on pretty shaky ground.
Thanks, that’s quite clear. Should I reference abandonment of fundamental objects as the major feature of a paradigm shift?
In fact, in every case of inexpressibility that we know of, it’s been because one of the ways of thinking about the world didn’t give correct predictions.
Yes, every successful paradigm shift. Proponents of failed paradigm shifts are usually called cranks. :)
My position is that the repeated pattern of false fundamental objects suggest that we should give up on the idea of fundamental objects, and simply try to make more accurate predictions without asserting anything else about the “accuracy” of our models.
Predictions about the world are only possible to the extent the world controls the predictions, to the extent considerations you use to come up with the predictions correspond to the state of the world. So it’s not possible to make useful predictions based on considerations that don’t correspond to reality, or conversely if you manage to make useful predictions, there must be something in your considerations that corresponds to the world. See Searching for Bayes-Structure.
Isn’t “makes accurate predictions” synonymous with “corresponds to reality in some way” ? If there was absolutely no correspondence between your model and reality, you wouldn’t be able to judge how accurate your predictions were. In order to make such a judgement, you need to compare your predictions to the actual outcome. By doing so, you are establishing a correspondence between your model and reality.
That is, the concept cannot be coherently expressed using the new terminology. For example, there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus.
I’m not seeing how the second sentence is an example of the criterion in your first sentence. That criterion seems to strict, too: in general the new paradigm subsumes the old (as in the canonical example of Newtonian vs relativistic physics).
I’m also not seeing what the attributes “empirical” and “non-social” have to do (causally) with the ability to form coherent concepts.
Maybe you should also unpack what you mean by “coherent”?
I’m not a mathematician, but from my outside perspective I would cheerfully qualify something like Wilf-Zeilberger theory as the math equivalent to a paradigm shift in the empirical sciences.
WP lists “non-euclidean geometry” as a paradigm shift, BTW.
That is, the concept cannot be coherently expressed using the new terminology. For example, there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus.
I’m not seeing how the second sentence is an example of the criterion in your first sentence.
Using modern physics, there is no way to express the concept that Ptolomy intended when he said epicycles. More casually, modern physicists would say “Epicycles don’t exist” But contrast, the concept of set is still used in Cantor’s sense, even though his formulation contained a paradox. So I think the move from geocentric theory to heliocentric theory is a paradigm shift, but adjusting the definition of set is not.
I’m also not seeing what the attributes “empirical” and “non-social” have to do (causally) with the ability to form coherent concepts.
I’m using the word science as synonymous with “empirical studies” (as opposed to making stuff up without looking). That’s not intended to be controversial in this community. What is controversial is the assertion that studying the history of science shows examples of paradigm shifts.
One possible explanation of this phenomena is that science is socially mediated (i.e. affected by social factors when the effect is not justified by empirical facts).
I’m asserting that mathematics is not based on empirical facts. Therefore, one would expect that it could avoid being socially mediated by avoiding interacting with reality (that is, I think a sufficiently intelligent Cartesian skeptic could generate all of mathematics). IF I am correct that they are caused by the socially mediated aspects of the scientific discipline and IF mathematics can avoid being socially mediated by virtue of its non-empirical nature, then I would expect that no paradigm shifts would occur.
This whole reference to paradigm shifts is an attempt to show a justification for my belief that mathematics is non-empirical, contrary to the original quote. If you don’t believe in paradigm shifts (as Kuhn meant them, not as used by management gurus), then this is not a particularly persuasive argument.
WP lists “non-euclidean geometry” as a paradigm shift, BTW.
If Wikipedia says that, I don’t think it is using the word the way Kuhn did.
WP lists “non-euclidean geometry” as a paradigm shift, BTW.
If Wikipedia says that, I don’t think it is using the word the way Kuhn did.
For Kuhn, the word was, if anything, a sociological term—not something referring to the structure of reality itself. (Kuhn was not himself a postmodernist; he still believed in physical reality, as distinct from human constructs.) So it seems to me that it would be entirely consistent with his usage to talk about paradigm shifts in mathematics, since the same kind of sociological phenomena occur in the latter discipline (even if you believe that the nature of mathematical reality itself is different from that of physical reality).
Using modern physics, there is no way to express the concept that Ptolomy intended when he said epicycles.
As I’d mentioned elsewhere, there’s actually a pretty easy way to express that, IMO: “Ptolemy thought that planets move in epicycles, and he was wrong for the following reasons, but if we had poor instruments like he did, we might have made the same mistake”.
IF I am correct that they are caused by the socially mediated aspects of the scientific discipline and IF mathematics can avoid being socially mediated by virtue of its non-empirical nature, then I would expect that no paradigm shifts would occur.
The abovementioned non-euclidean geometry is one such shift, as far as I understand (though I’m not a mathematician). I’m not sure what the difference is between the history of this concept, and what Kuhn meant.
But there were other, more powerful paradigm shifts in math, IMO. For example, the invention of (or discovery of, depending on your philosophy) zero (or, more specifically, a positional system for representing numbers). Irrational numbers. Imaginary numbers. Infinite sets. Calculus (contrast with Zeno’s Paradox). The list goes on.
I should also point out that many, if not all, of these discoveries (or “inventions”) either arose as a solution to a scientific problem (f.ex. Calculus), or were found to have a useful scientific application after the fact (f.ex. imaginary numbers). How can this be, if mathematics is entirely “non-empirical” ?
Hmm, I’ll have to think about the derivation of zero, the irrational numbers, etc.
I should also point out that many, if not all, of these discoveries (or “inventions”) either arose as a solution to a scientific problem (f.ex. Calculus), or were found to have a useful scientific application after the fact (f.ex. imaginary numbers). How can this be, if mathematics is entirely “non-empirical”
The motivation for derivation of mathematical facts is different from the ability to derive them. I don’t why the Cartesian skeptic would want to invent calculus. I’m only saying it would be possible. It wouldn’t be possible if mathematics was not independent of empirical facts (because the Cartesian skeptic is isolated from all empirical facts except the skeptic’s own existence).
I don’t why the Cartesian skeptic would want to invent calculus. I’m only saying it would be possible.
My point is that we humans are not ideal Cartesian skeptics. We live in a universe which, at the very least, appears to be largely independent of our minds (though of course our minds are parts of it). And in this universe, a vast majority of mathematical concepts have practical applications. Some were invented with applications in mind, while others were found to have such applications after their discovery. How could this be, if math is entirely non-empirical ? That is, how do you explain the fact that math is so useful to science and engineering ?
My position, FWIW, is that all of science is socially mediated (as a consequence of being a human activity), mathematics no less than any other science. Whether a mathematical proposition will be assessed as true by mathematicians is a property ultimately based on physics—currently the physics of our brains.
For example, there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus.
I disagree, as, I suspect, you already know :-)
But I have a further disagreement with your last sentence:
Because mathematics is non-empirical and therefore non-social...
What do you mean, “and therefore” ? As I see it, “empirical” is the opposite of “social”. Gravity exists regardless of whether I like it or not, and regardless of how many passionate essays I write about Man’s inherent freedom to fly by will alone.
Yes, non-empirical is the wrong word. I mean to assert that mathematics is independent of empirical fact (and therefore non-social. A sufficiently intelligent Cartesian skeptic could derive all of mathematics in solitude).
Goedel showed that no one can derive all of mathematics at all, whether in solitude or in a group, because any consistent system of axioms can’t lead to all the true statements from their domain.
Anyone know whether it’s proven that there are guaranteed to be non-self-referential truths which can’t be derived from a given axiom system? (I’m not sure whether “self-referential” can be well-defined.)
A sufficiently intelligent Cartesian skeptic could derive all of mathematics in solitude...
I don’t know whether this is true or not; arguments could (and have) been made that such a skeptic could not exist in a non-empirical void. But that’s a bit offtopic, as I still have a problem with your previous sentence:
I mean to assert that mathematics is independent of empirical fact … and therefore non-social.
Are you asserting that all things which are “dependent on empirical fact” are “social” ? In this case, you must be using the word “social” in a different way than I am.
If we lived in a culture where belief in will-powered flight was the norm, and where everyone agreed that willing yourself to fly was really awesome and practically a moral imperative… then people would still plunge to their deaths upon stepping off of skyscraper roofs.
I don’t know whether this is true or not; arguments could (and have) been made that such a skeptic could not exist in a non-empirical void.
:) It is the case that the coherence of the idea of the Cartesian skeptic is basically what we are debating.
I’m specifically asserting that things that are independent of empirical facts are non-social.
I think that things that are subject to empirical fact are actually subject to social mediation, but that isn’t a consequence of my previous statement.
What does rejection of the assertion “If you think you can fly, then you can” have to do with the definition of socially mediated? I don’t think post-modern thinking is committed to the anti-physical realism position, even if it probably should endorse the anti-physical models position. The ability to make accurate predictions doesn’t require a model that corresponds with reality.
It is the case that the coherence of the idea of the Cartesian skeptic is basically what we are debating.
That might be a bit orthogonal to the discussion; I’m certainly willing to grant you the Cartesian skeptic for the duration of this thread :-)
I’m specifically asserting that things that are independent of empirical facts are non-social.
If you are talking about pure reason, don’t the conclusions depend on your axioms ? If so, the results may not be social, per se, but they’re certainly arbitrary. If you pick different axioms, you get different conclusions.
What does rejection of the assertion “If you think you can fly, then you can” have to do with the definition of socially mediated? … The ability to make accurate predictions doesn’t require a model that corresponds with reality.
To me, these two sentences sound diametrically opposed to each other. If your model does not correspond to reality, how is it different from any other arbitrary social construct (such as the color of Harry Potter’s favorite scarf or whatever) ? On the other hand, if your model makes specific predictions about reality, which are found to be true time and time again (f.ex., “if you step off this ledge, you’ll plummet to your splattery doom”), then how can you say that your model does not correspond to reality in any meaningful way ?
The frequentist vs. baysian debate is a debate of computing mathematical paradigms. True mathematicians however shun statistics. They don’t like the statistical pradigm ;)
Gödel’s discovery ended a certain mathmatical pradigm of wanting to construct a complete mathematics from the ground up.
I could imagine a future paradigm shift way from the ideal of mathmatical proofs to more experimental math. Neural nets or quantum computers can give you answer to mathematical question that you ask that might be better than the answer s that axiom and proof based math provides.
I don’t think Deutsch means that mathematical proofs are all inductive. I think he means that proofs are constructed and checked on physical computing devices like brains or GPGPUs; and that because of that mathematical knowledge is not in a different ontological category than empirical knowledge.
I feel quite confident saying that mathematics will never undergo paradigm shifts, to use the terminology of Kuhn.
The same is not true for empirical sciences. Paradigm shifts have happened, and I expect them to happen in the future.
It believe it already has. Consider the Weierstrass revolution. Before Weierstrass, it was commonly accepted that while continuous functions may lack a derivative at a set of discrete points, it still had to have a derivative somewhere. Then Weierstrass developed a counterexample, which I think satisfies the Kuhnian “anomaly that cannot be explained within the current paradigm.”
Another quick example: during the pre-War period, most differential geometry was concerned with embedded submanifolds in Euclidean space. However, this formulation made it difficult to describe or classify surfaces—I seem to believe but don’t have time to verify that even deciding whether two sets of algebraic equations determine isomorphic varieties is NP-hard. Hence, in the post-War period, intrinsic properties and descriptions.
EDIT: I was wrong, or at least imprecise. Isomorphism of varieties can be decided with Grobner bases, the reduction of which is still doubly-exponential in time, as far as I can tell. Complexity classes aren’t in my domain; I shouldn’t have said anything about them without looking it up. :(
Reading the wiki page, it looks like Weierstrass corrected an error in the definition or understanding of limits. But mathematicians did not abandon the concept of limit the way physicists abandoned the concept of epicycle, so I’m not sure that qualifies as a paradigm shift. But I’m not mathematician, so my understanding may be seriously incomplete.
I can’t even address your other example due to my failure of mathematical understanding.
Hindsight bias. The old limit definition was not widely considered either incorrect or incomplete.
They abandoned reasoning about limits informally, which was de rigeur beforehand. For examples of this, see Weierstrass’ counterexample to the Dirichlet principle. Prior to Weierstrass, some people believed that the Dirichlet principle was true because approximate solutions exist in all natural examples, and therefore the limit of approximate solutions will be a true solution.
Not true. The “old limit definition” was non-existent beyond the intuitive notion of limit, and people were fully aware that this was not a satisfactory situation.
We need to clarify what time period we’re talking about. I’m not aware of anyone in the generation of Newton/Leibniz and the second generation (e.g., Daniel Bernoulli and Euler) who felt that way, but it’s not as if I’ve read everything these people ever wrote.
The earliest criticism I’m aware of is Berkeley in 1734, but he wasn’t a mathematician. As for mathematicians, the earliest I’m aware of is Lagrange in 1797.
I’m also curious about this history.
That’s pretty clear, thanks. Obviously, experts aren’t likely to think there is a basic error before it has been identified, but I’m not in position to have a reliable opinion on whether I’m suffering from hindsight bias.
Still, what fundamental object did mathematics abandon after Weierstrass’ counter-example? How is this different from the changes to the definition of set provoked by Russell’s paradox?
I don’t recall where it is said that such an object is necessary for a Kuhnian revolution to have occurred. There was a crisis, in the Kuhnian sense, when the old understanding of limit (perhaps labeling it as limit1 will be clearer) could not explain the existence of e.g., continuous functions without derivatives anywhere, or counterexamples to the Dirichlet principle. Then Weierstrass developed limit2 with deltas and epsilons. Limit1 was then abandoned in favor of limit2.
Wikipedia gives the acceptance of non-Euclidean geometry as a “classical case” of a paradigm shift. I suspect that there were several other paradigm shifts involved from Euclid’s math to our math: for instance, coordinate geometry, or the use of number theory applied to abstract quantities as opposed to lengths of line segments.
Would the whole Russel’s paradox incident count as a mathematical paradigm shift?
Reading Wikipedia, it looks like a naive definition of a set turns out to be internally inconsistent. Does that mean the concept of set was abandoned by mathematicians the way epicyles have been abandoned by physicists? That’s not my sense, so I hesitate to say redefining set in a more coherent way is a paradigm shift. But I’m no mathematician.
Its a matter of degree rather than an absolute line. However, I would say a time when even the very highest experts in a field believed something of great importance to their field with quite high confidence, and then turned out to wrong, probably counts.
I don’t think “everyone in field X made an error” is that same thing as saying “Field X underwent a paradigm shift.”
Why not ? That sounds like a massive shift in the core beliefs of the field in question. If that’s not a paradigm shift, then what is ?
The “non-expressible in the new concept-space” thing that you think never actually happens.
This looks very like trying to define away something that sure felt like a paradigm shift to the people in the field. Remember that “paradigm” is a belief held by people, not a property inherent in the universe.
Perhaps this is a limitation of my understanding of Kuhn, in that I’m misusing his terminology. I am unaware of mathematics abandoning fundamental objects as inherently misguided the way physics abandoned epicycles or impetus. I expect physics will have similar abandonments in the future, but I expect mathematics never will. The difference is a property of the difference between mathematics and empirical facts. This comment makes the argument I’m trying to assert in slightly different form.
Isn’t that exactly what happened? The phrase “set of all sets that do not contain themselves” isn’t really expressible in Zermelo-Fraenkel set theory, since that has a more limited selection of ways to construct new sets and “the set of everything that satisfies property X” is not one of them.
I don’t think it’s terribly useful to frame the discussion in terms of concepts that never actually happen :-)
What would count as one?
As I understand it, a paradigm shift would include the abandonment of a concept. That is, the concept cannot be coherently expressed using the new terminology. For example, there’s no way to express coherent concepts in things like Ptolomy’s epicycles or Aristole’s impetus. I think Kuhn would say that these examples are evidence that empirical science is socially mediated.
I’m not aware of any formerly prominent mathematical concepts that can’t even be articulated with modern concepts. Because mathematics is non-empirical and therefore non-social, I would be surprised if they existed.
A totally trivial nit pick, I admit, but there’s no such thing as the Aristotelian theory of impetus. The theory of impetus was an anti-Aristotelian theory developed in the middle ages. Aristotle has no real dynamical theory.
Thanks. Did not know that.
Thanks, I did not actually know that. But I should have known.
There are perfectly fine ways to express those things. Epicycles might even be useful in some cases, since they can be used as a simple approximation of what’s going on.
The reason people don’t use epicycles any more isn’t because they’re unthinkable, in the really strong “science is totally culture-dependent” sense. It’s because using them was dependent on whether we thought they reflected the structure of the universe, and now we don’t. Ptolemy’s claim behind using epicycles was that circles were awesome, so it was likely that the universe ran on circles. This is a fact that could be tested by looking at the complexity of describing the universe with circles vs. ellipses.
So this paradigm shift stuff doesn’t look very unique to me. It just looks like the refutation of an idea that happened to be central to using a model. Then you might say that math can have no paradigm shifts because it constructs no models of the world. But this isn’t quite true—there are models of the mathematical world that mathematicians construct that occasionally get shaken up.
My point was that trying to express epicycles in the new terminology is not possible. That is, modern physicists say, “Epicycles don’t exist.”
Obviously, it is possible to use sociological terminology to describe epicycles. You yourself said that they were useful at times. But that’s not the language of physics.
Since you mentioned it, I would endorse “Science is substantially culturally dependent”, NOT “Science is totally culturally dependent.” So culturally dependent that there is not reason to expect correspondence between any model and reality. Better science makes better predictions, but it’s not clear what a “better” model would be if there’s no correspondence with reality.
I brought all this up not to advocate for the cultural dependence of science. Rather, I think it would be surprising for a discipline independent of empirical facts to have paradigm shifts. Thus, the absence of paradigm shifts is a reason to think that mathematics is independent of empirical facts.
If you don’t think science is substantially culturally dependent, then there’s no reason my argument should persuade you that mathematics is independent of empirical facts.
This is false in an amusing way: expressing motion in terms of epicycles is mathematically equivalent to decomposing functions into Fourier series—a central concept in both physics and mathematics since the nineteenth century.
To be perfectly fair, AFAIK Ptolemy thought in terms of a finite (and small) number of epicycles, not an infinite series.
And so for the curves in question, the Fourier expansion would have only a finite number of terms.
The point being that, in contrast to what was being asserted, Ptolemy’s concept is subsumed within the modern one; the modern language is more general, capable of expressing not only Ptolemy’s thoughts, but also a heck of a lot more. In effect, modern mathematical physics uses epicycles even more than Ptolemy ever dreamed.
That’s good point; I haven’t thought about that. Go epicycles ! Epicycles to the limit !
ducks and runs away
But it is! You simply specify the position as a function of time and you’ve done it! The reason why that seems so strange isn’t because modern physics has erased our ability to add circles together, it’s because we no longer have epicycles as a fundamental object in our model of the world.
So if you want the copernican revolution to be a paradigm shift, the idea needs to be extended a bit. I think the best way is to redefine paradigm shift as a change in the language that we describe the world in. If we used to model planets in terms of epicycles, and now we model them in terms of ellipses, that’s a change of language, even though ellipses can be expressed as sums of epicycles, and vice versa.
In fact, in every case of inexpressibility that we know of, it’s been because one of the ways of thinking about the world didn’t give correct predictions. We have yet to find two ways of thinking about the world that let you get different experimental results if you plan the experiment two different ways. In these cases, the paradigm shift included the falsification of a key claim.
I don’t think it’s necessarily true (for example, you can imagine an abstract game having a revolution in how people thought about what it was doing), but it seems reasonable for math, depending on how you define “math.” I think people are just giving you a hard time because you’re trying to make this general definitional argument (generally not worth the effort) on pretty shaky ground.
Thanks, that’s quite clear. Should I reference abandonment of fundamental objects as the major feature of a paradigm shift?
Yes, every successful paradigm shift. Proponents of failed paradigm shifts are usually called cranks. :)
My position is that the repeated pattern of false fundamental objects suggest that we should give up on the idea of fundamental objects, and simply try to make more accurate predictions without asserting anything else about the “accuracy” of our models.
How can you make accurate predictions while at the same time discarding the notion of accuracy ?
I have no reason to expect that our models correspond to reality in any meaningful way, but I still think that useful predictions are possible.
Predictions about the world are only possible to the extent the world controls the predictions, to the extent considerations you use to come up with the predictions correspond to the state of the world. So it’s not possible to make useful predictions based on considerations that don’t correspond to reality, or conversely if you manage to make useful predictions, there must be something in your considerations that corresponds to the world. See Searching for Bayes-Structure.
Isn’t “makes accurate predictions” synonymous with “corresponds to reality in some way” ? If there was absolutely no correspondence between your model and reality, you wouldn’t be able to judge how accurate your predictions were. In order to make such a judgement, you need to compare your predictions to the actual outcome. By doing so, you are establishing a correspondence between your model and reality.
I’m not seeing how the second sentence is an example of the criterion in your first sentence. That criterion seems to strict, too: in general the new paradigm subsumes the old (as in the canonical example of Newtonian vs relativistic physics).
I’m also not seeing what the attributes “empirical” and “non-social” have to do (causally) with the ability to form coherent concepts.
Maybe you should also unpack what you mean by “coherent”?
I’m not a mathematician, but from my outside perspective I would cheerfully qualify something like Wilf-Zeilberger theory as the math equivalent to a paradigm shift in the empirical sciences.
WP lists “non-euclidean geometry” as a paradigm shift, BTW.
Using modern physics, there is no way to express the concept that Ptolomy intended when he said epicycles. More casually, modern physicists would say “Epicycles don’t exist” But contrast, the concept of set is still used in Cantor’s sense, even though his formulation contained a paradox. So I think the move from geocentric theory to heliocentric theory is a paradigm shift, but adjusting the definition of set is not.
I’m using the word science as synonymous with “empirical studies” (as opposed to making stuff up without looking). That’s not intended to be controversial in this community. What is controversial is the assertion that studying the history of science shows examples of paradigm shifts.
One possible explanation of this phenomena is that science is socially mediated (i.e. affected by social factors when the effect is not justified by empirical facts).
I’m asserting that mathematics is not based on empirical facts. Therefore, one would expect that it could avoid being socially mediated by avoiding interacting with reality (that is, I think a sufficiently intelligent Cartesian skeptic could generate all of mathematics). IF I am correct that they are caused by the socially mediated aspects of the scientific discipline and IF mathematics can avoid being socially mediated by virtue of its non-empirical nature, then I would expect that no paradigm shifts would occur.
This whole reference to paradigm shifts is an attempt to show a justification for my belief that mathematics is non-empirical, contrary to the original quote. If you don’t believe in paradigm shifts (as Kuhn meant them, not as used by management gurus), then this is not a particularly persuasive argument.
If Wikipedia says that, I don’t think it is using the word the way Kuhn did.
For Kuhn, the word was, if anything, a sociological term—not something referring to the structure of reality itself. (Kuhn was not himself a postmodernist; he still believed in physical reality, as distinct from human constructs.) So it seems to me that it would be entirely consistent with his usage to talk about paradigm shifts in mathematics, since the same kind of sociological phenomena occur in the latter discipline (even if you believe that the nature of mathematical reality itself is different from that of physical reality).
As I’d mentioned elsewhere, there’s actually a pretty easy way to express that, IMO: “Ptolemy thought that planets move in epicycles, and he was wrong for the following reasons, but if we had poor instruments like he did, we might have made the same mistake”.
The abovementioned non-euclidean geometry is one such shift, as far as I understand (though I’m not a mathematician). I’m not sure what the difference is between the history of this concept, and what Kuhn meant.
But there were other, more powerful paradigm shifts in math, IMO. For example, the invention of (or discovery of, depending on your philosophy) zero (or, more specifically, a positional system for representing numbers). Irrational numbers. Imaginary numbers. Infinite sets. Calculus (contrast with Zeno’s Paradox). The list goes on.
I should also point out that many, if not all, of these discoveries (or “inventions”) either arose as a solution to a scientific problem (f.ex. Calculus), or were found to have a useful scientific application after the fact (f.ex. imaginary numbers). How can this be, if mathematics is entirely “non-empirical” ?
Hmm, I’ll have to think about the derivation of zero, the irrational numbers, etc.
The motivation for derivation of mathematical facts is different from the ability to derive them. I don’t why the Cartesian skeptic would want to invent calculus. I’m only saying it would be possible. It wouldn’t be possible if mathematics was not independent of empirical facts (because the Cartesian skeptic is isolated from all empirical facts except the skeptic’s own existence).
My point is that we humans are not ideal Cartesian skeptics. We live in a universe which, at the very least, appears to be largely independent of our minds (though of course our minds are parts of it). And in this universe, a vast majority of mathematical concepts have practical applications. Some were invented with applications in mind, while others were found to have such applications after their discovery. How could this be, if math is entirely non-empirical ? That is, how do you explain the fact that math is so useful to science and engineering ?
Hmm, “justified” generally has a social component, so I doubt that this definition is useful.
So this WP page doesn’t exist? ;)
My position, FWIW, is that all of science is socially mediated (as a consequence of being a human activity), mathematics no less than any other science. Whether a mathematical proposition will be assessed as true by mathematicians is a property ultimately based on physics—currently the physics of our brains.
I disagree, as, I suspect, you already know :-)
But I have a further disagreement with your last sentence:
What do you mean, “and therefore” ? As I see it, “empirical” is the opposite of “social”. Gravity exists regardless of whether I like it or not, and regardless of how many passionate essays I write about Man’s inherent freedom to fly by will alone.
Yes, non-empirical is the wrong word. I mean to assert that mathematics is independent of empirical fact (and therefore non-social. A sufficiently intelligent Cartesian skeptic could derive all of mathematics in solitude).
Didn’t Gödel show that nobody can derive all of mathematics in solitude because you can’t have a complete and consistented mathamatical framework?
Goedel showed that no one can derive all of mathematics at all, whether in solitude or in a group, because any consistent system of axioms can’t lead to all the true statements from their domain.
Anyone know whether it’s proven that there are guaranteed to be non-self-referential truths which can’t be derived from a given axiom system? (I’m not sure whether “self-referential” can be well-defined.)
It is. At least, it’s possible to express Goedel statements in the form “there exist integers that satisfy this equation”.
It can’t.
I don’t know whether this is true or not; arguments could (and have) been made that such a skeptic could not exist in a non-empirical void. But that’s a bit offtopic, as I still have a problem with your previous sentence:
Are you asserting that all things which are “dependent on empirical fact” are “social” ? In this case, you must be using the word “social” in a different way than I am.
If we lived in a culture where belief in will-powered flight was the norm, and where everyone agreed that willing yourself to fly was really awesome and practically a moral imperative… then people would still plunge to their deaths upon stepping off of skyscraper roofs.
:) It is the case that the coherence of the idea of the Cartesian skeptic is basically what we are debating.
I’m specifically asserting that things that are independent of empirical facts are non-social.
I think that things that are subject to empirical fact are actually subject to social mediation, but that isn’t a consequence of my previous statement.
What does rejection of the assertion “If you think you can fly, then you can” have to do with the definition of socially mediated? I don’t think post-modern thinking is committed to the anti-physical realism position, even if it probably should endorse the anti-physical models position. The ability to make accurate predictions doesn’t require a model that corresponds with reality.
That might be a bit orthogonal to the discussion; I’m certainly willing to grant you the Cartesian skeptic for the duration of this thread :-)
If you are talking about pure reason, don’t the conclusions depend on your axioms ? If so, the results may not be social, per se, but they’re certainly arbitrary. If you pick different axioms, you get different conclusions.
To me, these two sentences sound diametrically opposed to each other. If your model does not correspond to reality, how is it different from any other arbitrary social construct (such as the color of Harry Potter’s favorite scarf or whatever) ? On the other hand, if your model makes specific predictions about reality, which are found to be true time and time again (f.ex., “if you step off this ledge, you’ll plummet to your splattery doom”), then how can you say that your model does not correspond to reality in any meaningful way ?
The frequentist vs. baysian debate is a debate of computing mathematical paradigms. True mathematicians however shun statistics. They don’t like the statistical pradigm ;)
Gödel’s discovery ended a certain mathmatical pradigm of wanting to construct a complete mathematics from the ground up.
I could imagine a future paradigm shift way from the ideal of mathmatical proofs to more experimental math. Neural nets or quantum computers can give you answer to mathematical question that you ask that might be better than the answer s that axiom and proof based math provides.
Except, in practice mathematics still works this way.