home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:17577 comp.theory:2778 sci.logic:2526
- Path: sparky!uunet!think.com!hsdndev!husc-news.harvard.edu!scws7.harvard.edu!zeleny
- From: zeleny@scws7.harvard.edu (Michael Zeleny)
- Newsgroups: sci.math,comp.theory,sci.logic
- Subject: Self-Reference and Coding
- Message-ID: <1993Jan1.122745.18924@husc3.harvard.edu>
- Date: 1 Jan 93 17:27:44 GMT
- Organization: The Phallogocentric Cabal
- Lines: 60
- Nntp-Posting-Host: scws7.harvard.edu
-
- Consider a formula in the language of Robinson Arithmetic. Ask the
- question whether it is self-referential, in the sense of containing a
- term which encodes it via some scheme of G\"odel numbering. To what
- extent can this question be answered, effectively or otherwise?
-
- Motivation: assume a compositional intensional semantics, i.e. an
- ontology of concepts for all constants and formulae. The fundamental
- rule is that the concept of a formula is an appropriate function on
- the concepts of its terms. Example: for $1+1=2$ (pardon my TeX, --
- the dollar sign is a math delimiter), have the concept $1_1 +_1 1_1
- =_1 2_1$, where the subscript 1 indicates the first concept of each
- thing, whether individual, function, or relation. Agree to omit
- subscripts from function and relation constants, -- they can be
- restored from the context. We say that terms and formulae express
- their concepts; the concept of a formula is also called a proposition.
-
- Define a relation of synonymy, which partitions the concepts into
- equivalence classes, and induces a Lindenbaum algebra of the
- corresponding sentences, to be that of logical equivalence. (This is
- Church's Alternative (2); another notion of synonymy is given by the
- Alternative (1), which takes \lambda-convertibility as its definition.
- Since the language is typed, the latter relation is decidable.) So,
- for instance, $1_1+1_1$ is synonymous with $2_1$ as concepts of the
- integer 2. Unfortunately, this means that the synonymy relation is
- undecidable, inasmuch as our language may allow us to express the same
- concept via such rocky roads as the Goldbach conjecture or Fermat's
- last theorem.
-
- Now, to exclude G\"odel-type sentences, I stipulate that a necessary
- condition for a concept to be meaningful, is that it not contain any
- well-formed constituents synonymous with itself. Note that this is
- precisely the intensional analogue of the set-theoretic Axiom of
- Foundation, which ensures that all meaningful concepts may be assigned
- a rank in the cumulative hierarchy of concepts. Thus, in particular,
- the sentences $\psi$, for which $$\psi \equiv \neg \mbox{Pr}_{\cal
- Q}(\lceil \psi \rceil),$$ (incidentally, what is the proper AMS LaTeX
- symbol for corner quotes?) that is, such that they are materially
- equivalent to the sentence denying the provability in Robinson
- arithmetic of the sentence with their G\"odel number, are deemed
- meaningless. Since nothing depends on the negation, the L\"ob
- sentences, which assert their own provability, are likewise excluded.
- All meaningless sentences (i.e. ones not expressing a meaningful
- concept) are regarded as bereft of truth-values. All formal proofs
- are required to be valid, i.e to contain only meaningful formulae.
- Now say that synonymy is *provable* logical equivalence, and
- reconstrue the above accordingly.
-
- In accordance with the above considerations, this means that
- meaningfulness is not an effectively decidable notion. (Gerald Sacks
- says that Church tried what sounds like a similar approach towards
- analyzing G\"odel's results. I am unaware of any published work, so
- references would be greatly appreciated.) However, it appears to be
- semi-decidable. What else can be said about this notion?
-
- Happy New Year to all.
-
- cordially,
- mikhail zeleny@husc.harvard.edu
- "Les beaulx bastisseurs nouveaulx de pierres mortes ne sont escriptz
- en mon livre de vie. Je ne bastis que pierres vives: ce sont hommes."
-