On 18 Nov. Steven Ericsson-Zenith wrote:
My own interpretation may be substantively different, it may not. I take Hilbert's position to be that the formalism is independent of the subject matter. That is, I take his view of formal interpretation to be mechanistic, specifying valid transformations of the structure under consideration, be it logical, geometric or physical. I am confused because you use "signs" instead of "marks" here. In addition, since the formalism is independent of the subject - as suggested by his appeal to Berkeley - a theorem of the formalism remains a theorem of the formalism despite the subject. In this view, how one selects an appropriate formalism for a given subject - if there is a fitness ("suitability") requirement as you suggest for the "different parts of mathematics" - appears to be a mystery, unless you think empiricism is required at this point. And on 26 Nov., Jerry Chandler wrote: The separate and distinct axiom systems for mathematical structures is a thorn in my mind as it disrupts simpler notions of the rules for conducting calculations with numbers. While I eventually came to accept the category theorists view of the emergence of mathematic structures as a historical fact, the separation of formal axiom systems causes philosophical problems. Firstly, physicists often speak of "first principles" or "ab initio" foundations. These terms are used in such a sense as to imply a special connection exists between physics and the universals and to further imply that other sciences do not have access to such "first principles". If such ab initio calculations were to be invoked as something more serious than a linguistic fabrication, what mathematical structure would one invoke? In the mid and late 20 th Century, group theory and symmetry were the popular choice among philosophically oriented physicists and applied mathematicians. Philosophically, the foundations of Aristotelian causality come into play. Philosophers abandoned material causality, substituted the formalism of efficient causality for formal causality and summarily dissed telic reasoning of biology. The four causes so widely discussed in medieval logic and the trivium were reduced to one cause. Obviously, if every axiom system is a formalism and every formalism is to be associated with efficient causality, the concept of "ab initio" calculations is seriously diminished as every axiom system becomes a new form of "ab initio" calculations and conclusions. Does anyone else see this as a problem for the philosophy of physics? A second question is perhaps easier for you, Irving, or perhaps more challenging. You write:
... the only mathematically legitimate characteristic of axiom systems is that they be proof-theoretically sound, that is, the completeness, consistency, and independence of the axiom system,...
I am puzzled on how to interpret the phrase,
and independence of the axiom system,...
Many axiom systems may use the same axioms, just extend the number of axioms in the system; the formal axiom systems are interdependent on one another. So, what is the sense of 'independence' as it is used in this phrase? I would note in passing that I have attempted to write a set of axioms for chemical relations on several occasions and have been amply rewarded, initially with exhilaration and subsequently with deep remorse for having wasted my energies on such an intractable challenge. :-) :-) :-) I would like to take the easiest question first. Independence, like consistency or completeness, is a model-theoretic property of axiom systems. To put it in simplest terms, by the independence of the axiom system, the mathematician means nothing more nor less than that there are no axioms in the set of axioms that could be proven as a theorem from the other axioms. The issue arose when, ever since Euclid, mathematicians attempted to determine the status of the parallel postulate. It came to a head in 1733 when Saccheri claimed to prove the parallel postulate (Vth) from all of Euclid's axioms other than the Vth postulate, using a reductio argument. What Beltrami showed in 1868 was that Saccheri had in fact proven the independence of Euclid's Vth postulate, since in fact, Saccheri ended up with a hyperbolic parallel postulate. I don't know very much about Hilbert's axiomatization of physics, other than that Leo Corry has written the most about them, including the book David Hilbert and the Axiomatization of Physics (1898-1918) (Dordrecht: Kluwer Academic Publishers, 2004); some of his other work includes: "David Hilbert and the Axiomatization of Physics", Archive for History of Exact Sciences 51 (1997), 83-198; "Hilbert and Physics (1900-1915)," In Jeremy Gray (ed,), The Symbolic Universe: Geometry and Physics (18901930) (New York, Oxford University Press, 1999), 145187; "On the Origins of Hilbert's 6th Problem: Physics and the Empiricist Approach to Axiomatization", in Marta Sanz-Solé et al (eds.), Proceedings of the International Congress of Mathematicians, Madrid 2006, (Zurich, European Mathematical Society, 2006), vol. 3, 1679-1718. These, and many of Corry's other publications, are downloadable from his web page at: http://www.tau.ac.il/~corry/publications/hilbert.html Speaking of Corry, I would have to say that he would agree that there is a strong empiricism underlying Hilbert's work, and that this is the philosophical import of his quote from Kant's K.d.r.V. in the Grundlagen der Geometrie: "So fängt denn alle menschliche Erkenntnis mit Amschauungen an, geht von da zu Begriffen und endigt mit Ideen." I would argue, however, that this is about how we obtain our information, and, assuming Corry is correct, how Hilbert thought we select the elements of our universe of discourse; but I would also argue that it has nothing to do with how axiomatic systems operate, which is to say, having established the axioms, chosen the inference rules for the system, and selected the primitives from which theorems are constructed from the axioms in accordance with the inference rules, is strictly mechanical, and it does not, working within the axiom system, whether what is being manipulated are points, lines and planes, or tables, chairs, and beer mugs, or integers, , or whatever we may require for the axiomatizing task at hand. What matters within the system, while the calculations are occurring, is that complex formulas (theorems) are being constructed on the basis of the formulas that do duty as axioms, in accordance with the rules. (It is this distinction, of having inference rules in place, that renders Hilbert's systems not merely axiomatic systems, but formal deductive systems.) Hilbert's formalism amounts to the mechanization of these manipulations, and for practical purposes, the formulas are combinations of marks, and these marks become signs as soon as an interpretation is give, that is, a universe of discourse - - whether points, lines, and planes, or tables, chairs and beer mugs, or the integers. What concerns me is whether, in considering what (else) or what different Hilbert might have meant by his formalism, and whether or not there was an underlying empiricism behind this, is that we might be demanding too much of Hilbert, who was, I understand, concerned with mathematics and only peripherally with philosophy of mathematics. (Having said this, I have to also confess that I have not seen or read the contents of his late, unpublished, lectures on foundations, but I believe that Corry has, and it is on that basis that Corry proposes an empiricist epistemology behind Hilbert's formalism.) The only other point I would make w.r.t. Hilbert on physics, is that, at least according to Corry, part of Hilbert's empiricism is exhibited by the requirement that his axiomatization depends upon his axiomatization of geometry, and that the Kantian root of geometry is spatial intuition. While selecting axioms is definitely nontrivial, I hardly find it to be a mystery. When I look at a piece of mathematics, I have a conception of what its properties are, and what behavior to expect. (Practically speaking, when developing a piece of mathematics, the working mathematician starts with what is already known and has an idea already in mind for the theorem to be added. In effect, before writing a proof, the mathematician is working backward, from the theorem to the axioms that are required, to which are added the theorems already proven. This is not being flippant here. In fact, the new field of reverse mathematics that emerged in proof theory in the work of Stephen Simpson and Harvey Friedman starting in the mid 1970s does precisely that - - attempt to determine, from the theorem(s) required for a particular piece of mathematics, what the minimal axiom set would be to enable proof of the theorem(s) in question.) We know that the more structure a mathematical object has, the more axioms it requires. Consider the following example. Suppose we start with the axioms of ZF (most mathematicians' favorite axiomatization), with or without AC. If we add to a set an associative binary operation and has an identity element I in S such that for all a in S, we get the next level of complexity, namely monoids. To go from a monoid to a groupoid, we require invertibility. The properties of a group are closure, associativity, the identity property, and the inverse property. Therefore, a group is a set, finite or infinite, whose axioms guarantee these properties. If we are dealing with additive groups, we make certain that our group operator is group addition and be careful that association, the inverse element and the identity element are chosen for addition; if multiplicative groups, mutatis mutandis, being careful that the group operator is group multiplication, and that association the inverse element, the identity element, are defined for multiplication. One more axiom, commutativity, gives us abelian groups. It is also worth noting that being careful in selecting the primitives for the universe of discourse can have a significant influencr on the size of one's axiom set as well. For example, by starting with the real line to measure lengths and angles, Birkhoff found a shorter list minimal list of axioms for his treatment of geometry than Hilbert's, and R. L. Moore in 1902 demonstrated that Hilbert's Axiom II.4 was redundant (for details on the redundancy, see William Richter's paper, at http://www.math.northwestern.edu/~richter/hilbert.pdf I do not know how to understand the relevance for this discussion of causality, Aristotelian or otherwise. Does causality have some to do with the inference rules? This is probably an issue better left to philosophers. The closest I came to dealing with this - - perhaps - - is a term paper (no longer extant) on "What Hume Might Have Said to Peano", probably back in 1970 or 1971, in which I noted that Hume, in the Treatise on Human Nature, argued that a relation is required to pass from the addition of two units to the number two, whereas Peano did not include among, or along with, his axioms for natural numbers, a rule or relation that would allow him to proceed to 2 from 1 and 1 taken together. When I wrote this paper, I did not yet own a copy of van Heijenoort's From Frege to Gödel, in which, as we know, van Heijenoort asserted that Peano's axiomatic system did not contain any inference rule that enabled passage from 1 + 1 to 2, as a consequence Peano devised an axiomatic system, but not a formal deductive system. Does the presence or absence of an inference rule mean the difference between a formal deductive system and a [mere] axiomatic system? Is there some sort of causality, Aristotelian or otherwise, in [application of] inference rules? Once again, I am at a loss here to comprehend how this issue of causality relates to the nature of axiom systems or to formalism. Irving Irving H. Anellis Visiting Research Associate Peirce Edition, Institute for American Thought 902 W. New York St. Indiana University-Purdue University at Indianapolis Indianapolis, IN 46202-5159 USA URL: http://www.irvinganellis.info --------------------------------------------------------------------------------- You are receiving this message because you are subscribed to the PEIRCE-L listserv. To remove yourself from this list, send a message to lists...@listserv.iupui.edu with the line "SIGNOFF PEIRCE-L" in the body of the message. To post a message to the list, send it to PEIRCE-L@LISTSERV.IUPUI.EDU