- #1

- 2,255

- 1

- Thread starter pivoxa15
- Start date

- #1

- 2,255

- 1

- #2

- 252

- 1

I'd argue that nowadays there are way fewer the mathematicians who are supporters of mathematical platonism (I include here also Popper's view that mathematical sentences are our own creation but they exist independent of us in a world of culture, his 'Third World')...Formalism about numbers is far more popular at least from Hilbert onward. Sure Hilbert's program does not work till the end (Godel's legacy) but the formalist programe is still tenable to some extent. In this view syntactically correct formulas which are undecidable in a certain set of symbols/axioms/rules of inference are not true/false by default (some say 'true/false but indemonstrable' about undecidable statements). It is us which make them true/false either by adding such statements (or their contraries) as new axioms or by modifying the initial set of axioms so that these undecidable initially formulations appear as theorems in the new system (that is they are deducible from the new set of axioms).pivoxa15 said:

Last edited:

- #3

- 491

- 0

Godel told us that there are statements about the integers which cannot be proved, but which we nevertheless

Of course the situation is never simple. Penrose is well known as a Platonist, but he seems to steer away from the questions of infinite sets, prefering to base his ideas on a more geometric viewpoint.

- #4

- 252

- 1

The formalist could use Russell's definition of a number (classes of classes) and argue that all we need to define let's say natural numbers is an intensional description though the extension is not listable (extension can always be reduced to intension, the reverse is not always valid). In this view natural numbers are not seen as Platonic, the empirical correspondence with phenomenal facts (observed facts) of some undecidable statements being contingent.

Last edited:

- #5

Hurkyl

Staff Emeritus

Science Advisor

Gold Member

- 14,916

- 19

Let me clarify: in anyGodel told us that there are statements about the integers which cannot be proved, but which we nevertheless know to be true.

It's not as impressive as it sounds (though it's still an impressive result): Gödel's theorem proves that there exist statements in the theory of the integers that cannot be proven or disproven from the axioms of the integers. Since one of the things a model does is that it gives a way to assign to each statement a truth value, we must have a true statement that cannot be proven from the axioms of the integers.

- #6

- 491

- 0

When I talk about "the integers", I mean the "standard model", 1, 2, 3 ... My point is that in the end there is no way to formally define this, you have to have some Platonist notion of the (standard model of the) integers being "out there"Hurkyl said:Let me clarify: in anymodelof the integers, there exists true statements that cannot be proved from the axiom of the integers.

- #7

- 252

- 1

I think we must make a clear distinction between the truth (of mathematical statements) as formalists understand it (true mathematical statements are axioms or can be deduced from them) and truth understood as conformity with some empirical facts.

Formalists hold that mathematical statements have no meanings, we must attach a specific semantics when use mathematics in physical sciences. So that a mathematical statement can be undecidable in a certain system of axioms of mathematics and still to be 'true' in relation with empirical facts as much as we attach the required semantics. Schrodinger's equation is a good example here (it was 'invented' by Schordinger using an analogy with classical physics).

If we add Born's intepretation of the wavefunction we can say that it is 'true' in relation with empirical facts (it is part of the system of axioms used by science, held, at least provisionally, as true) though it is undecidable in the system of axioms of mathematics (at least it was so in the 1920s). I think the problem of numbers falls in the same case, there is no necessary link between the two types of truths. All we need to talk of mathematical truth is an intensional definition of numbers within formalism, otherwise we should endorse Kant's view that there is apriori synthetic knowledge (not the best of options in my view).

Formalists hold that mathematical statements have no meanings, we must attach a specific semantics when use mathematics in physical sciences. So that a mathematical statement can be undecidable in a certain system of axioms of mathematics and still to be 'true' in relation with empirical facts as much as we attach the required semantics. Schrodinger's equation is a good example here (it was 'invented' by Schordinger using an analogy with classical physics).

If we add Born's intepretation of the wavefunction we can say that it is 'true' in relation with empirical facts (it is part of the system of axioms used by science, held, at least provisionally, as true) though it is undecidable in the system of axioms of mathematics (at least it was so in the 1920s). I think the problem of numbers falls in the same case, there is no necessary link between the two types of truths. All we need to talk of mathematical truth is an intensional definition of numbers within formalism, otherwise we should endorse Kant's view that there is apriori synthetic knowledge (not the best of options in my view).

Last edited:

- #8

honestrosewater

Gold Member

- 2,105

- 5

As for the OP, you can search for formalists and logicists. They include Hilbert, Carnap, Tarski, Curry, Frege, and many others. I'm not sure about intuitionist and constructivists - some may agree with platonism in ways.

- #9

Hurkyl

Staff Emeritus

Science Advisor

Gold Member

- 14,916

- 19

That's wrong. In Zermelo set theory, it goes like this:My point is that in the end there is no way to formally define this,

(Recall that in the usual model set-theoretic of the natural numbers, 0 is modelled by the empty set, and the successor operation is modelled by x → x U {x})

Then we define

In group theory, one could say:

In ring theory, one could say:

In topos theory, it goes like this:

A natural number object is an object N and arrows 1-->N and N-->N (called "zero" and "successor") such that, for any other pair of arrows 1 --> A --> A, we have a unique commutative diagram:

Code:

```
1 --> N --> N
| | |
| | |
V V V
1 --> A --> A
```

(This diagram encodes the idea of recursively defining a function N --> A)

- #10

- 491

- 0

No, I don't buy that.Hurkyl said:That's wrong. In Zermelo set theory, it goes like this:

(Recall that in the usual model set-theoretic of the natural numbers, 0 is modelled by the empty set, and the successor operation is modelled by x → x U {x})

Axiom of Infinity:There exists a set S such that [itex]\emptyset \in S[/itex] and [itex]x \in S \implies x \cup \{ x \} \in S[/itex].

Then we defineNto be the smallest set having that property. (It takes a slightly roundabout method to actually say "the smallest set having that property", though)

If you just use first order axioms then you won't be able to uniquely characterize the "standard model" of the integers (Godel tells us that there will be nonstandard models of your axioms)

And higher order axioms will correspond to a Platonist notion of actually inifinite sets or the like.

For instance in the real number axioms the real numbers (and the integers within them) are defined uniquely, but the completeness axiom is 2nd order, requiring arbitrary infinite sequences of rationals.

- #11

AKG

Science Advisor

Homework Helper

- 2,565

- 4

There are a couple of questions here, however, that do depend on philosophical beliefs. 2+2=4 may be true for everyone, but is "2" the number a real thing or "just a concept"? Is that even a true dilemma, or can something that is a concept be real? For some people, a number is only real if there is something in physical reality that occurs in that quantity. There being 2 apples is, for some people, proof that 2 is real. So what exactly are number (and other mathematical objects), are they real, and in what sense?honestrosewater said:

As for the OP, you can search for formalists and logicists. They include Hilbert, Carnap, Tarski, Curry, Frege, and many others. I'm not sure about intuitionist and constructivists - some may agree with platonism in ways.

The other question is that of mathematical propositions. Are they purely man-made, and hence is their truth contingent, or are they necessary truths? In other words, is the "world" of mathematical propositions created and contingent on the mind of man, or is it simply accessed by our minds while it exists independently?

- #12

Hurkyl

Staff Emeritus

Science Advisor

Gold Member

- 14,916

- 19

I don't see why the possibility of nonstandard models matters either, to

By the way, when cast in the language of sets, the completeness axiom is a

[tex]

\forall S, T \subseteq \mathbb{R}:

(\forall s \in S, t \in T: s \leq t) \implies

\exists x \in \mathbb{R}: (\forall s \in S, t \in T: s \leq x \leq t)

[/tex]

The corresponding formulation in topos theory would also be a first-order statement.

On the other hand, if you're working in the language of ordered fields, it is a second order statement (because S and T range over first-order unary relations on

The proof is by Tarski's theorem: For any particular S and T, the specialization of the completeness axiom is a first-order statement, and is true in the reals, therefore it's true in every real closed field. Since S and T were arbitrary, the conclusion follows.

- #13

mathwonk

Science Advisor

Homework Helper

- 11,041

- 1,232

- #14

selfAdjoint

Staff Emeritus

Gold Member

Dearly Missed

- 6,786

- 7

So true! Mathematical talent sometimes seems more like what the Ancient Greeks thought of Eros: a cruel and irresponsible god who takes you over willy-nilly and gives you deep personal rewards while ruining your social life.mathwonk said:

- #15

- 2,255

- 1

- #16

vanesch

Staff Emeritus

Science Advisor

Gold Member

- 5,028

- 16

I think the formalist dream (Hilbert's programme) has turned out not to work. Sure, For All Practical Purposes (FAPP), it is still a useful tool, but it has its limits (Goedel).pivoxa15 said:Even Hilbert showed signs of it when he praised Cantor's work so highly. How can one be both a formalist and Platonist at the same time?

And in fact that shouldn't surprise us. After all, formal logic is nothing else but a game of finite sequences of a finite set of symbols that satisfy certain formal rules. Formal logic is the study of a certain set of mappings which map {1,2,...n} into, say, the ASCII set. It only has a meaning when we GIVE it a meaning - and that's Platonism, no ?

I'd even say, for this formal game to have a meaning in the first place, you'd already have to ASSUME certain properties of natural numbers, in order to be able to say things about these mappings from {1...n} into the ASCII set. So using then this formal game to PROVE properties of natural numbers seems to be a bit circular IMHO, because you've USED certain properties of the natural numbers to set up the formal machinery in the first place.

For instance, how do you prove that every natural number has a successor, purely formally ?

- #17

honestrosewater

Gold Member

- 2,105

- 5

Well, don't take this too seriously - I haven't given it much thought. Take, say, the (finite number of) symbols of English (alphabet, punctuation symbols, etc.). Imagine all possible strings of these symbols, say, 36 symbols long. Some of the strings will be grammatical English sentences (or words, phrases, etc.), even though I didn't include any rules of English grammar in the system. This will be one of those sentences. Is that not a pure coincidence? If not, what would make it a pure coincidence?vanesch said:I think the formalist dream (Hilbert's programme) has turned out not to work. Sure, For All Practical Purposes (FAPP), it is still a useful tool, but it has its limits (Goedel).

And in fact that shouldn't surprise us. After all, formal logic is nothing else but a game of finite sequences of a finite set of symbols that satisfy certain formal rules. Formal logic is the study of a certain set of mappings which map {1,2,...n} into, say, the ASCII set. It only has a meaning when we GIVE it a meaning - and that's Platonism, no ?

I'd even say, for this formal game to have a meaning in the first place, you'd already have to ASSUME certain properties of natural numbers, in order to be able to say things about these mappings from {1...n} into the ASCII set. So using then this formal game to PROVE properties of natural numbers seems to be a bit circular IMHO, because you've USED certain properties of the natural numbers to set up the formal machinery in the first place.

For instance, how do you prove that every natural number has a successor, purely formally ?

Why can't you think of a theory as just having been chosen from all the possible options? - And in what way, if any, would the choice not be arbitrary?

- #18

matt grime

Science Advisor

Homework Helper

- 9,395

- 3

do you have any evidence the "most well known mathematicians" are platonists? Can you even name me any non-dead ones? Of course whatever you views on what maths is in this sense don't affect your ability to do it, and I would hazard a guess that in a world post Goedel, in which the continuum hypothesis and the axiom of choice are independent of the axioms of ZF, in which it is possible to construct a theoretical game in which both players always win, where the Banach Tarski paradox holds, then we are more likely to be formalists of some description, I know I am (not that i am at all a "good pure mathematician"). Indeed I only know of one other mathematician's declared view, and he (Gowers, Field's Medal) is certainly not a platonist.pivoxa15 said:

http://www.dpmms.cam.ac.uk/~wtg10/philosophy.html

Last edited:

- #19

selfAdjoint

Staff Emeritus

Gold Member

Dearly Missed

- 6,786

- 7

Roger Penrose is a declared Platonist. And Goedel was one too, according to Rebecca Goldstein's new book about him,matt grime said:do you have any evidence the "most well known mathematicians" are platonists? Can you even name me any non-dead ones? Of course whatever you views on what maths is in this sense don't affect your ability to do it, and I would hazard a guess that in a world post Goedel, in which the continuum hypothesis and the axiom of choice are independent of the axioms of ZF, in which it is possible to construct a theoretical game in which both players always win, where the Banach Tarski paradox holds, then we are more likely to be formalists of some description, I know I am. Indeed I only konw of one other mathematician's declared view, and he (Gowers, Field's Medal) is certainly not a platonist.

http://www.dpmms.cam.ac.uk/~wtg10/philosophy.html

- #20

matt grime

Science Advisor

Homework Helper

- 9,395

- 3

Apart from Penrose do we have any living declared PLatonists out there? (Hawking is allegedly a positivist if that is a counter balance to Penrose.) Bear in mind that the nature of mathematics has changed dramatcially over the last 100 years.

In fact let me slightly rephrase it as: apart from Penrose who is 'famously' a platonist, and Gowers who just happens to have written a talk about philosophy and mathematics where he reveals he isn't a platonist, has any modern 'big' mathematician actually ever said what they think? I guess 'no' and this underlines the fact that whatever your view of it, if you even have a view, doesn't matter one jot. Although I would put myself as a formalist of some kind I can't say I have ever thought about it in anyway other than to decide, after some prompting like this, that I do not consider there to be any reason for platonism to be 'true', indeed you might make a case that it is less that I am a formalist than simply "not a platonist".

In fact let me slightly rephrase it as: apart from Penrose who is 'famously' a platonist, and Gowers who just happens to have written a talk about philosophy and mathematics where he reveals he isn't a platonist, has any modern 'big' mathematician actually ever said what they think? I guess 'no' and this underlines the fact that whatever your view of it, if you even have a view, doesn't matter one jot. Although I would put myself as a formalist of some kind I can't say I have ever thought about it in anyway other than to decide, after some prompting like this, that I do not consider there to be any reason for platonism to be 'true', indeed you might make a case that it is less that I am a formalist than simply "not a platonist".

Last edited:

- Replies
- 26

- Views
- 915

- Last Post

- Replies
- 10

- Views
- 2K

- Replies
- 13

- Views
- 2K

- Replies
- 4

- Views
- 1K

- Replies
- 6

- Views
- 944

- Last Post

- Replies
- 22

- Views
- 2K

- Replies
- 17

- Views
- 1K

- Last Post

- Replies
- 8

- Views
- 3K

- Replies
- 3

- Views
- 2K

- Replies
- 12

- Views
- 2K