This is a ``preproof'' accepted article for The Review of Symbolic Logic.

This version may be subject to change during the production process.

DOI: 10.1017/S1755020321000526

Gödel’s Theorem and Direct Self-Reference

Saul A. Kripke

Abstract
In his paper on the incompleteness theorems, Gödel seemed to say
that a direct way of constructing a formula that says of itself that it is
unprovable might involve a faulty circularity. In this note, it is proved
that ‘direct’ self-reference can actually be used to prove his result.
Keywords/MSC2010 Classification Codes: 03-03, 03B10, direct self-
reference, Gödel numbering, Incompleteness.

In his epochal 1931 paper on the incompleteness theorems Gödel wrote,

regarding his undecidable statement, saying, in effect, “I am unprovable,”

Contrary to appearances, such a proposition involves no faulty

circularity, for initially it [only] asserts that a certain well-defined
formula (namely, the one obtained from the qth formula in the
lexicographic order by a certain substitution) is unprovable. Only
subsequently (and so to speak by chance) does it turn out that
this formula is precisely the one by which the proposition itself
was expressed.[1, p. 151, fn. 15]

The tendency of this footnote might be that if one tried to express the state-
ment directly, there might indeed be a faulty circularity. In my own paper,
“Outline of a Theory of Truth”, I expressed some doubt about this ten-
dency, and stated that the Gödel theorem could be proved using ‘direct’
self-reference. To quote this paper,

A simpler, and more direct, form of self-reference uses demon-

stratives or proper names: Let ‘Jack’ be a name of the sentence
‘Jack is short’, and we have a sentence that says of itself that it
is short. I can see nothing wrong with “direct” self-reference of
this type. If ‘Jack’ is not already a name in the language,4 why
can we not introduce it as a name of any entity we please? In
particular, why can it not be a name of the (uninterpreted) finite
sequence of marks ‘Jack is short’ ? (Would it be permissible to call

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526
 this sequence of marks “Harry,” but not “Jack”? Surely prohibi-
tions on naming are arbitrary here.) There is no vicious circle in
our procedure, since we need not interpret the sequence of marks
‘Jack is short’ before we name it. Yet if we name it “Jack,” it at
once becomes meaningful and true. (Note that I am speaking of
self-referential sentences, not self-referential propositions.5 )
In a longer version, I would buttress the conclusion of the preced-
ing paragraph not only by a more detailed philosophical exposi-
tion, but also by a mathematical demonstration that the simple
kind of self-reference exemplified by the “Jack is short” example
could actually be used to prove the Gödel incompleteness theorem
itself (and also, the Gödel-Tarski theorem on the undefinability
of truth). Such a presentation of the proof of the Gödel theo-
rem might be more perspicuous to the beginner than is the usual
one. It also dispels the impression that Gödel was forced to re-
place direct self-reference by a more circumlocutory device. The
argument must be omitted from this outline.6 [6, pp. 77–78].
4
We assume that ‘is short’ is already in the language.
5
It is not obviously possible to apply this technique to obtain
“directly” self-referential propositions.
6
There are several ways of doing it, using either a nonstandard
Gödel numbering where statements can contain numerals desig-
nating their own Gödel numbers, or a standard Gödel numbering,
plus added constants of the type of ‘Jack’.

As I say in footnote 6, there are several ways of obtaining the Gödel theorem
using direct self-reference, analogously to ‘Jack is short’. Note that, as I say
in footnote 5, I am not claiming that this technique could be used to obtain
self-referential propositions. Perhaps this is what Gödel had in mind in his
own footnote. Let us look at how Gödel’s first incompleteness theorem is to
be done these ways.
One way, observed independently by Raymond Smullyan, is to use a non-
standard Gödel numbering where a formula can contain a numeral desig-
nating its own Gödel number.1 This is clearly impossible under the usual
Gödel prime power numbering, or various other variants. That this should
not happen might be a natural restriction, since a Gödel number might be
thought to correspond with a formula as a composite object, and this might
1
Almost everything I say about Smullyan is based on conversation with him. We
haven’t been able to find a published reference.

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526
 be thought to preclude a formula from containing a numeral designating its
own Gödel number.
Nevertheless we propose a nonstandard Gödel numbering allowing a state-
ment to contain a numeral designating its own Gödel number. We can pro-
ceed as follows. Let x1 be a fixed variable. Let A1 (x1 ), A2 (x1 ), . . . be an
enumeration of all those formulae that contain at most x1 free (Gödel’s class
signs). We assume that the language of the system studied contains, either
directly or by virtue of interpretation, the language of arithmetic. Numerals
can be assumed to be terms 0 followed by (or preceded by) finitely many suc-
cessor symbols (allowing none). We use 0(n) for 0 with n successor symbols;
n = 0 is allowed, as I said. 0(n) denotes the number n.
Let the ‘original’ Gödel numbering be Gödel’s own prime power product
numbering, except that the smallest prime used is 3, so that Gödel numbers
are always odd. In the ‘new’ numbering, all Gödel numbers coincide with
the ‘original’, except that for each n, the formula

(∃x1 )(x1 = 0(2kn ) ∧ An (x1 ))

gets the Gödel number 2kn , where kn is the original Gödel number of (∃x1 )(x1 =
0(n) ∧ An (x1 )). The ‘new’ numbering allows a formula to contain a numeral
designating its Gödel numbering, and in that sense it is a self-referential
Gödel numbering.
In this self-referential Gödel numbering, every formula An (x1 ) has an
‘instance’ (∃x1 )(x1 = 0(2kn ) ∧ An (x1 )) asserting that its own Gödel number
satisfies An (x1 ). The Gödel incompleteness theorem is the special case where
An (x1 ) is unprovability in the system.
The ‘new’ Gödel number is in effective 1 − 1 correspondence with the old
one, where the inverse function is also effective. The complexity of a property
in the arithmetical hierarchy (Σ01 , Π01 , etc.) does not change when the ‘new’
Gödel numbers replace the old.2
Remark 1: We could have simply used An (0(2kn ) ). But using (∃x1 )(x1 =
(2kn )
0 ∧An (x1 )) has two advantages, both emphasized by Raymond Smullyan.
First, concatenation is easier to arithmetize than substitution, which has
some value even here.3 Second, this choice guarantees that no formula gets
2
The existential quantifier in (∃x1 )(x1 = 0(2kn ) ∧ An (x1 )) does not increase complexity,
since the formula is equivalent to A(0(2kn ) ). For the Gödel theorem, the important feature
is that ‘unprovability in the system’ is still a Π01 predicate. If the language contains
non-arithmetical predicates that have hierarchical status, say, in the hyperarithmetical or
analytical hierarchy, these also remain the same.
Further, since the domain and range of both of these 1 − 1 mappings are recursive, the
mapping can be extended to a recursive permutation of the natural numbers.
3
The second reason is the most important one. I have not worried about this in the

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526
 two Gödel numbers, which is not clear if we use the simpler version. It also
may not be harmful if some formulae get two or more Gödel numbers, but in
that case we cannot think of formulae as identified with their Gödel numbers.
Remark 2: A referee has proposed an alternative and simpler ‘new’ Gödel
numbering. For each n, and formula A(x1 ), let (∃x1 )(x1 = 0(2n) ∧ An (x1 )) get
the Gödel number 2n. Otherwise, the ‘old’ (always odd) numbers are used.
Much more natural, in my opinion, for getting ‘direct’ self-reference, is
the use of added constants.4 Once again, let A1 (x1 ), A2 (x1 ), . . . , Ai (x1 ) be
an enumeration of all those formulae of the language L containing at most
the variable x1 free. Let a1 , a2 , . . . be a denumerable list of constants, none
of which are in the original language L. Give Gödel numbers to the formulae
of the extended language in a conventional way. Now add to the axioms of
the original system S, an infinite set of axioms ai = 0(ni ) , where ni is the
Gödel number of Ai (ai ). The resulting system is S 0 . Pretty clearly S 0 is a
conservative extension of S. It simply extends S by adding an infinite set of
constants with specific numerical values—new names for particular numbers.
Every proof in S 0 becomes a proof in S if each constant is replaced by the
corresponding numeral.
Now, for the Gödel (first incompleteness) theorem, consider the predicate
(in the original language, without the extra constants) that says that x1 is
the Gödel number of a formula unprovable in the extended system S 0 . Call
this ∼ ThmS 0 (x1 ). This is a Π01 predicate and is a formula Ai (x1 ) for some
i. In our construction, there is a constant ai of S 0 such that ai = 0(ni ) is
an axiom of S 0 , where ni is the Gödel number of Ai (ai ). Then it is easily
shown, in the usual way, that Ai (ai ) is unprovable in S 0 if S 0 is consistent, and
that ∼ Ai (ai ) is unprovable in S 0 if S 0 is ω-consistent (or even 1-consistent).
Given that S 0 is a conservative extension of S, the same things must be true
of Ai (0(ni ) ) in the original system S as well.
The particular case of unprovability is just one example. For any definable
property we obviously can use this construction to formulate a statement that
‘says of itself’ that it has that property (the ‘self-reference lemma’).
case of added constants. But substitution of a constant for a fixed free variable x1 is in
fact easy to arithmetize provided x1 does not occur bound in A(x1 )—in other words, that
‘(x1 )’ does not occur in it.
4
One of the referees has commented, regarding numerals, “it is highly plausible to view
them as proper names; but they have also been seen as analogues of quotational names and
structural-descriptive names. Numerals—unlike constants—are complex closed terms and
at least in this respect similar to a term built from a term with function symbols for the
primitive recursive functions for substitution and numerals”. The doubts in question do
not apply to constants, as the referee says. Nevertheless, the use of a self-referential Gödel
numbering, though somewhat artificial, does seem more directly self-referential than the
traditional Gödelian argument.

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526
 A more delicate construction uses only a single constant a, and a single
extra axiom, in defining the system S ∗ extending S. It will have only one
extra axiom, which will be called (∗). In the case of Gödel’s first incom-
pleteness theorem, we wish to construct a statement saying that it itself is
unprovable. So given the constant a, we consider the single formula standing
for the statement
a is unprovable in the system S ∗ , which is S extended
(∗)
by adding pa = 0(a) q as a single axiom

(Of course this is to be written out symbolically.) The constant a occurs in

(∗) only as explicitly shown. Note that in (∗), the constant a is both used
and mentioned. In a standard Gödel numbering of the extended language,
(∗) has a particular Gödel number. Say it is n. Then let the extended system
S ∗ be axiomatized by adding a = 0(n) as an axiom.
Given this extra axiom, a = 0(n) , note that (∗) will say that it itself (i.e.,
its Gödel number) is unprovable in the system S ∗ itself. Now we can once
again deduce that (∗) is unprovable in S ∗ if S ∗ is consistent, its negation is
unprovable in S ∗ if S ∗ is ω-consistent, etc. Also once again S ∗ is a conser-
vative extension of S, so that any proof in S is valid in S ∗ iff it was valid in
S, just as in the previous argument with an infinity of constants. And once
again, this allows us to transfer the result to the original system S. Clearly
also the argument extends to results about any definable A(x1 ) in place of
unprovability (the self-reference lemma).
Remark. In spite of my own use of the Quinean terminology of ‘use’ and
‘mention’ in the preceding paragraph, I am not following his practices of
quotation and quasi-quotation. Many formal expressions in the paper are
used autonymously. The corners in (∗) are used to denote Gödel numbers of
formulae pa = 0(a) q, where a is a constant on the left, but on the right it is an
as yet unspecified numeral (the constant a is used, rather than mentioned).
In other words, ‘0(a) ’ is not a term in the language, since it means 0 followed
by a successors, and therefore what formula is mentioned in the corners is
yet to be specified. In the whole formula in corners, a is ‘mentioned’.5

Acknowledgments My thanks to Yale Weiss for editorial support. Special

thanks to Romina Padró, Panu Raatikainen, and Albert Visser for their
comments and for encouraging me to write this paper. This paper has been
completed with support from the Saul Kripke Center, at The City University
of New York, Graduate Center.
5
A referee has suggested that the reader would profit from the following further readings
on direct self-reference: [5, 8, 2, 3, 4, 7].

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526
 Affiliation Distinguished Professor in the Philosophy and Computer Sci-
ence Programs at The Graduate Center, CUNY. Address The Saul Kripke
Center, The Graduate Center, CUNY, 365 Fifth Ave., Room 7118, New
York, NY 10016, USA. Email [email protected].

References
[1] Gödel, K. “On formally undecidable propositions of Principia Mathemat-
ica and related systems I.” Translated by Jean van Heijenoort. In Feferman,
S., et al. (eds). Kurt Gödel: Collected Works, Volume I (pp. 145–195). New
York: Oxford University Press, 1986.

[2] Grabmayr, B., & Visser, A. “Self-reference upfront: A study of self-

referential Gödel numberings.” arXiv:2006.12178 [math.LO] (2020).

[3] Halbach, V., & Visser, A. “Self-reference in arithmetic I.” Review of Sym-
bolic Logic 7 (2014a), 671–691.

[4] Halbach, V., & Visser, A. “Self-reference in arithmetic II.” Review of

Symbolic Logic 7 (2014b), 692–712.

[5] Heck, R. “Self-reference and the languages of arithmetic.” Philosophia

Mathematica 15 (2004), 1–29.

[6] Kripke, S. “Outline of a theory of truth.” The Journal of Philosophy

72 (1975), 696–716. Reprinted in Kripke, S. Philosophical Troubles (pp.
75–98). New York: Oxford University Press, 2011.

[7] Picollo, L. “Reference in arithmetic.” Review of Symbolic Logic 11 (2018),

573–603.

[8] Visser, A. “Semantics and the liar paradox.” In Gabbay, D., & Guenthner,
F. (eds). Handbook of Philosophical Logic, Volume 11, second edition (pp.
149–240). Heidelberg: Springer, 2004.

Downloaded from https://www.cambridge.org/core. IP address: 98.221.247.81, on 03 Dec 2021 at 02:34:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1755020321000526