§1. I propose to address not so much Gödel's own philosophy of mathematics as the philosophical implications of his work, and especially of his incompleteness theorems. Now the phrase “philosophical implications of Gödel's theorem” suggests different things to different people. To professional logicians it may summon up thoughts of the impact of the incompleteness results on Hilbert's program. To the general public, if it calls up any thoughts at all, these are likely to be of the attempt by Lucas  and Penrose  to prove, if not the immortality of the soul, then at least the non-mechanical nature of mind. One goal of my present remarks will be simply to point out a significant connection between these two topics.
But let me consider each separately a bit first, starting with Hilbert. As is well known, though Brouwer's intuitionism was what provoked Hilbert's program, the real target of Hilbert's program was Kronecker's finitism, which had inspired objections to the Hilbert basis theorem early in Hilbert's career. (See the account in Reid .) But indeed Hilbert himself and his followers (and perhaps his opponents as well) did not initially perceive very clearly just how far Brouwer was willing go beyond anything that Kronecker would have accepted. Finitism being his target, Hilbert made it his aim to convince the finitist, for whom no mathematical statements more complex than universal generalizations whose every instance can be verified by computation are really meaningful, of the value of “meaningless” classical mathematics as an instrument for establishing such statements.
REALISM VS NOMINALISM
Philosophy is a subject in which there is very little agreement. This is so almost by definition, for if it happens that in some area of philosophy inquirers begin to achieve stable agreement about some substantial range of issues, straightaway one ceases to think of that area as part of “philosophy,” and begins to call it something else. This happened with physics or “natural philosophy” in the seventeenth century, and has happened with any number of other disciplines in the centuries since. Philosophy is left with whatever remains a matter of doubt and dispute.
Philosophy of mathematics, in particular, is an area where there are very profound disagreements. In this respect philosophy of mathematics is radically unlike mathematics itself, where there are today scarcely ever any controversies over the correctness of important results, once published in refereed journals. Some professional mathematicians are also amateur philosophers, and the best way for an observer to guess whether such persons are talking mathematics or philosophy on a given occasion is to look whether they are agreeing or disagreeing.
One major issue dividing philosophers of mathematics is that of the nature and existence of mathematical objects and entities, such as numbers, by which I will always mean positive integers 1, 2, 3, and so on. The problem arises because, though it is common to contrast matter and mind as if the two exhausted the possibilities, numbers do not fit comfortably into either the material or the mental category.
TWO SENSES OF “FOUNDATIONS OF MATHEMATICS”
Does mathematics requires a foundation? The first thing that must be said about the question is that the expression “foundations of mathematics” is ambiguous. Let me explain.
Modern mathematicians inherited from antiquity an ideal of rigor, according to which each mathematical theorem should be deduced from previously admitted results, and ultimately from an explicit list of postulates. It also inherited a further ideal according to which the postulates should be self-evidently true. During the great creative period of early modern mathematics, there were and probably had to be many departures from both ideals. But during the century before last, as mathematicians were driven or drawn to consider less familiar mathematical structures, from hyperbolic spaces to hypercomplex numbers, the need for rigor was increasingly felt, and higher standards were eventually instituted. But while the ideal of rigor may be claimed to have been realized, the ideal of selfevidence was not.
Considering only the ideal of rigor, the working mathematician's understanding of its requirements, of what is permissible in the way of modes of definition and modes of deduction of new mathematical notions and results from old, is largely implicit. Logic, which investigates such matters, and fixes explicit canons, is a subject in which the algebraist, analyst, or geometer need never take a formal course. Nor are mathematicians in practice much concerned with tracing back the chain of definitions and deductions beyond the recent literature in their fields to the ultimate primitives and postulates.
Some philosophers approach mathematics saying, “Here is a great and established branch of knowledge, encompassing even now a wonderfully large domain, and promising an unlimited extension in the future. How is mathematics, pure and applied, possible? From its answer to this question the worth of a philosophy may be judged.”
Other philosophers approach mathematics in a quite different spirit. They say, “Here is a body, already large and still being extended, of what purports to be knowledge. Is it knowledge, or is it delusion? Only philosophy and theology, from their standpoint prior and superior to that of mathematics and science, are worthy to judge.” While this inquisitorial conception of the relation between philosophy and science is less widely held today that it was in Cardinal Bellarmine's time, it continues to have many distinguished advocates.
Prominent among these is Michael Dummett, who has repeatedly advanced arguments for the claim that much of current mathematical theory is delusory and much of current mathematical practice is in need of revision – arguments for the repudiation, within mathematical reasoning, of the canons of classical logic in favor of those of intuitionistic logic. While nearly everything Dummett has written is pertinent in one way or another to his case for intuitionism, there are two texts especially devoted to stating that case: his much anthologized article (Dummett 1973a) on the philosophical basis of intuitionistic logic; and the concluding philosophical chapter of his guidebooks (Dummett 1977) to the elements of intuitionism.
Which if any of the many systems of modal logic in the literature is it whose theorems are all and only the right general laws of necessity? That depends on what kind of necessity is in question, so I should begin by making distinctions.
A first distinction that must be noted is between metaphysical necessity or inevitability – “what could not have been otherwise” – and logical necessity or tautology – “what it is self-contradictory to say is otherwise.” The stock example to distinguish the two is this: “Water is a compound and not an element.” Water could not have been anything other than what it is, a compound of hydrogen and oxygen; but there is no self-contradiction in saying, as was often said, that water is one of four elements along with earth and air and fire.
The logic of inevitability might be called mood logic, by analogy with tense logic. For the one aims to do for the distinction between the indicative “it is the case that…” and the subjunctive “it could have been the case that…,” something like what the other does for the distinction between the present “it is the case that…” and the future “it will be the case that…” or the past “it was the case that…” The logic of tautology might be called endometalogic, since it attempts to treat within the object language notions that classical logic treats only in the metalanguage.
Quine and his critique
Today there appears to be a widespread impression that W. V. Quine's notorious critique of modal logic, based on certain ideas about reference, has been successfully answered. As one writer put it some years ago: “His objections have been dead for a while, even though they have not yet been completely buried.” What is supposed to have killed off the critique? Some would cite the development of a new “possible-worlds” model theory for modal logics in the 1960s; others, the development of new “direct” theories of reference for names in the 1970s.
These developments do suggest that Quine's unfriendliness towards any formal logics but the classical, and indifference towards theories of reference for any singular terms but variables, were unfortunate. But in this study I will argue, first, that Quine's more specific criticisms of modal logic have not been refuted by either of the developments cited, and further, that there was much that those who did not share Quine's unfortunate attitudes might have learned about modality and about reference by attention to that critique when it first appeared, so that it was a misfortune for philosophical logic and philosophy of language that early reactions to it were as defensive and uncomprehending as they generally were.
“NOMINALISM” AND “REALISM”
Nominalism is a large subject. In our book (Burgess and Rosen 1997) my colleague Gideon Rosen and I distinguished a negative or destructive side of nominalism, which tells us not to believe what mathematics appears to say, from a positive or reconstructive side, which aims to give us something else to believe instead. We noted that there were a few nominalists who contented themselves with the negative side, conceding that mathematics is useful, insisting that what it appears to say is not true, and letting it go at that, without attempting any reconstrual or reconstruction of mathematics. We expressed some surprise that there were not more such destructive nominalists, since as compared with reconstructive nominalism, destructive nominalism has what Russell in another context called “the advantages of theft over honest toil”; and if nothing else was clear from the work of Hartry Field, Charles Chihara, Geoffrey Hellman, and other reconstructive nominalists whose work we surveyed, it was clear that the amount of honest toil that would be required for a nominalistic reconstrual or reconstruction of mathematics would be quite considerable.
Today, a couple of years after publication, it is beginning to seem that the main achievement of our book will have been to provide a decent burial for the hard-working, laborious variety of nominalism.
Responding to Harvey's theories about the circulation of the blood, Dr. Diafoirus argues (a) that no such theory was taught by Galen, and (b) that Harvey is not licensed to practice medicine in Paris. Plainly there is something wrong with a response of this sort, however effective it may prove to be in swaying an audience. For either or both of (a) and (b) might well be true without Harvey's theory being false. So Diafoirus's argument can serve only to divert discussion from the real question to irrelevant sideissues. The traditional term for such diversionary debating tactics is “fallacy of relevance.”
In recent years this tradition has come to be used in a quite untraditional sense among followers of N. D. Belnap, Jr., and the late A. R. Anderson. (All citations of these authors are from their masterwork Anderson and Belnap (1975), and are identified by page number.) According to these selfstyled “relevant logicians,” it is items (IA) and (IIA) in Table 13.1 that constitute the archetypal “fallacies of relevance.” (In the table ∼, &, and ⋁ stand for truth-functional negation, conjunction, and disjunction, respectively.) These forms of argument, say Anderson and Belnap, are “simple inferential mistake[s], such as only a dog would make” (p. 165). The authors can hardly find terms harsh enough for those who accept these schemata: they are called “perverse” (p. 5) and “psychotic” (p. 417).
Email your librarian or administrator to recommend adding this to your organisation's collection.