War Nez 2020
War Nez 2020
Mathematical Incompleteness
and Divine Ineffability
1. Introduction1
Perhaps ever since the revelation of the Divine Name on Sinai,
humans have acknowledged the power of self-referential definitions.
The ancient Hebrews so esteemed God’s self-description—“I am
who I am” (Ex 3:14)—that they eschewed its vocalization. Without
subjecting the name to a philosophical analysis, they intuited that it
captured something of divine ineffability.2
The type of self-reference of interest here is not produced by
merely any use of the first-person pronoun. Rather, we have in
mind self-referential statements which not only assign properties,
but which prescribe a definition. Of course, self-referential defini-
tions are no less powerful than they are hazardous. When used care-
lessly they run the risk of paradox, and Scripture itself knows of
this paradoxical dimension, especially in the form of the so-called
Epimenides paradox. Paul’s letter to Titus carries a notable example:
“One of [the Cretans] said, ‘Cretans are always liars’” (Ti 1:12). St.
Jerome also discovered the Epimenides paradox in Psalm 116:11,
where he explained the problem as follows:
l o g o s 23 :3 s u m m e r 2020
50 logos
“I said in my alarm, Every man is a liar!” Is David telling the
truth or is he lying? If it is true that every man is a liar, and Da-
vid’s statement, “Every man is a liar” is true, then David also is
lying; he, too, is a man. But if he, too, is lying, his statement:
“Every man is a liar,” consequently is not true. Whatever way
you turn the proposition, the conclusion is a contradiction.3
2. Mathematical Incompleteness
Rather than presenting a detailed exposition of the incompleteness
theorems here, I will attempt to bring nonmathematical readers into
contact with the main [Link] incompleteness theorems pertain to
formal systems, systems roughly comprising the following elements:
(1) a set of symbols, (2) a set of basic symbol-strings, called primitive
terms, (3) a set of rules for building formulas and sentences out of
primitive terms, (4) a set of unquestioned sentences, called axioms,
and (5) a set of inference rules. Any sentence that can be deduced by
applying the inference rules to combinations of the axioms is called
a theorem.
One can see why such a logical system would be called “formal,”
for it reduces logical reasoning to syntactic operations on a set of
symbols. The validity of a deductive procedure can be verified on
the basis of its syntactic “form,” and such evaluations can be made
52 logos
algorithmically, that is, without the aid of human ingenuity.
Formal systems can be used to “formalize” various branches of
mathematics, but here we shall focus on arithmetic. Arithmetical sys-
tems can be developed in different ways, but ultimately they provide
syntactic rules for deducing arithmetical theorems having an arbi-
trary degree of complexity. As long as we pin down the right axioms
and the right rules of inference, it’s conceivable that we could design
a computer program that would enumerate, one after the other, all
the theorems of arithmetic. The trouble is that standard formal sys-
tems of arithmetic have considerable expressive power—too much
for their own good, as we shall see. Gödel’s main achievement was
to show that standard systems strong enough to formalize arithmetic
are able to realize self-referential definitions, along with the atten-
dant hazards.
Gödel was able to get arithmetic to “talk about” itself by mapping
every theorem of arithmetic into a unique natural number. We’ll
indicate his procedure by creating an encoding for sentences of an
extremely simple formal system. Suppose our language contains the
equals sign (=) and glyphs for all the natural numbers (0, 1, 2, . . . ).
Then, let every sentence of our language take the form m = n where
m and n are natural numbers. To keep things simple, we won’t add
any axioms or inference rules to our formal system. Next we cre-
ate an encoding. Let the code number of the equals sign be 1, let
the code number of the symbol 0 be 2, let the code number of the
symbol 1 be 3, and so forth. Then we can let the code number of the
sentence m = n be given by 2m + 2 . 31 . 5n + 2. For example, the code
number of the sentence 1 = 3 is 75,000. In this way, we can translate
every sentence (whether true or false) into a unique natural num-
ber, and, by the unique prime factorization theorem, this process is
reversible.7 Of course, not all numbers decode into a well-formed
sentence.
Notice that in our above example, our language did not include
addition and multiplication (nor any of the logical symbols such as ¬,
∧, and ∃). Therefore, the encoding process (including the operation
mathematical incompleteness and divine ineffability 53
2m + 2 . 31 . 5n + 2) was not expressible in the language itself. Gödel,
however, was working with the full language of arithmetic, and there-
fore he could design an encoding process using that very language. He
thus was able to construct, for a certain formal system of arithmetic,
a self-referencing sentence G called the Gödel sentence. The sentence
G is defined to assert that G itself is not provable (within the formal
system, and according to the standard interpretation of arithmetic).8
Gödel’s first incompleteness theorem arises out of an inspection
of the Gödel sentence G. Suppose G is provable in the formal sys-
tem. Then, by the definition of G, G is not provable in that system.
So if G is provable, it is also not provable. Such a system, for which
a sentence and its negation are both provable, is called inconsistent.
But suppose we assume that our system is consistent. Then the only
remaining possibility is that G cannot be provable. But this is merely
what G asserts. So G is a true sentence (according to the standard in-
terpretation of arithmetic) which is not provable. A theory of arith-
metic that fails to prove all true sentences of arithmetic is called in-
complete. Thus, Gödel’s first incompleteness theorem runs as follows:
3. Physical Incompleteness
“Science is based on mathematics; mathematics cannot discover all
truths; therefore science cannot discover all truths.” This is how cos-
mologist John D. Barrow caricatures the argument that mathemati-
cal incompleteness limits scientific knowledge.13 More precisely, the
argument runs as follows: if there exists some kind of isomorphism
between a formal language of arithmetic and the laws of physics,
then the incompleteness theorems set epistemological constraints on
scientific explanation. In reality, the task of establishing an analogy
between scientific theories and formal systems is fraught with diffi-
culty, and there is hardly consensus on the practical consequences of
“physical incompleteness.” Nevertheless, some estimable physicists
suspect that the universe is so complex that any sufficiently detailed
description would harbor mathematical incompleteness.
Among those who subscribe to the physical incompleteness argu-
ment, Barrow identifies two schools of interpretation. The first is pes-
simistic, and is epitomized by Catholic philosopher of science Stanley
Jaki: “It seems that on the strength of Gödel’s theorem . . . the ultimate
foundations of . . . mathematical physics will remain embedded for-
ever in . . . the haziness of analogy and intuitions. For the speculative
physicist . . . there are limits to the precision of certainty [and] an
integral part of this boundary is the scientist himself, as a thinker.”14
For Jaki, the incompleteness theorems dash all hopes that science
might ultimately reach the deepest truths about the universe, let alone
metaphysical or spiritual truth. It seems that Jaki’s philosophical in-
tuition leads him to grant the incompleteness theorems a real-world
existence: something about a person writing down an exhaustive de-
scription of physics—physics by which that person is writing down the
description—seems doubtful, if not paradoxical. It would be like try-
ing to write one’s autobiography, without omitting the smallest detail.
56 logos
On the other hand, some physicists interpret physical incomplete-
ness more optimistically. As an example, Barrow points to Freeman
Dyson, for whom the incompleteness theorem is like “an insurance
policy against the scientific enterprise, which he admires so much,
coming to an end.”15 For Dyson, no matter the state of progress,
there will always be physical truths beyond the scope of our current
understanding, and so the thrill of discovery is interminable. Simi-
larly, Stephen Hawking is “glad that our search for understanding
will never come to an end,” thanks to the analogy between Gödel’s
theorems and quantum gravity.16
Whatever one makes of these two interpretations, both contain
aspects of the Christian worldview. Dyson’s and Hawking’s end-
lessly revelatory view of science finds parallels in heaven’s face-to-
face encounter with the infinite God. It is as if traces of the beatific
vision were built into the fabric of physics. On the other hand,
Jaki reminds us that physicists will always be left wanting insofar
as they seek heaven on earth, without proceeding through Christ
himself.
Still, many mathematicians, scientists, and philosophers are un-
willing to apply the incompleteness theorems to the natural world.
Gödel himself, though a friend of Albert Einstein and a student of
general relativity, was reluctant to mix physics with his incomplete-
ness results. He seemed to share Einstein’s distrust of quantum me-
chanics, and believed the universe to be thoroughly comprehensible,
at least in principle. In particular, Gödel refused to entertain the
notion that Heisenberg’s uncertainty principle might be the physical
embodiment of the incompleteness theorems.
While Gödel never defended himself on this point, Barrow pro-
vides a number of reasons why physical theories might be immune
to the incompleteness theorems. In the first place, argues Barrow,
the universe could be finite. The fundamental laws of physics might
refer to primitive entities that are, in fact, discrete. In such a case,
a theory of physics could be mathematically complete in principle,
mathematical incompleteness and divine ineffability 57
since Gödel’s theorems can only be derived in formal systems having
infinite possible terms.
Second, the incompleteness theorems are blocked from phys-
ics if, in fact, the laws of physics are not complex enough to per-
mit Gödel’s prime number encoding. The laws of physics, as pres-
ently understood, certainly contain sufficient arithmetic to produce
the incompleteness theorems. But here Barrow presents a striking
point: perhaps physical laws are, in reality, far simpler than they are
presently written.17 Consider the facts that (1) the incompleteness
theorems do not necessarily arise in geometry,18 and (2) Einstein
succeeded in giving a geometric foundation for gravitation. Thus, at
least in principle, the incompleteness theorems could be blocked in
general relativity, and perhaps the rest of physics could be given a
similarly geometric foundation. In other words, arithmetic might be
far more complicated than what is needed for a comprehensive de-
scription of the universe.
Thus, if the incompleteness theorems are not physically realiz-
able, the religious imagination is once again amused. Perhaps the
universe is finite—as Hebrew cosmology seems to suppose—so that
its intractability is an artifact of its astronomically large (yet finite)
number of constituents, and not some metaphysical stricture. Or
perhaps, if our most fundamental theories of physics are found to
admit the incompleteness theorems, it is simply because we have not
yet reached the most elegant expression of physical laws.
4. Cognitive Incompleteness
The “computational theory of mind” is the point of departure for
studying the human intellect in light of the incompleteness theo-
rems. According to this hypothesis, typically associated with a ma-
terialist worldview, the acts of human cognition are reducible to the
information processing methods of a sophisticated computer pro-
gram. This mind-emulating program, like any formal system capable
58 logos
of performing arithmetical operations, would be constrained by the
incompleteness theorems. Thus, if cognitive acts were purely com-
putational, the incompleteness theorems should also be found to
constrain the powers of the mind.
At least superficially, however, the incompleteness theorems
themselves run contrary to the computational theory of mind. For
any given mind-emulating program, the incompleteness theorems
allow us to generate a Gödel sentence, that is, an arithmetical truth
that the program is unable to prove by its own powers. But, by the
process of generating such a sentence, we (as humans) are able to see
that the Gödel sentence is true, even though the computer cannot.19
Apparently, then, the human mind has more cognitive power than
any computer program that is imagined to be its cognitive e quivalent.
In 1961, philosopher John R. Lucas exploited this ostensible par-
adox in order to launch an attack against the computational theory
of mind. Decades later, the argument was revisited and strengthened
by physicist Roger Penrose, and the so-called “Lucas–Penrose argu-
ment” continues to attract a great deal of scholarship. Here we will
only investigate some of the main issues, and to do so we will avail
ourselves to the syllogistic summary of Stephen Barr, rearranged for
our purposes.20 In Barr’s estimation, the Lucas–Penrose argument
runs as follows:
If the first three premises are granted, the human mind (in its
purely mathematical abilities) can surpass the limitations of the
mathematical incompleteness and divine ineffability 59
incompleteness theorems, and thus cannot be equivalent to a com-
puter program.
Not all Christian mathematicians have assented to the Lucas–
Penrose argument, although presumably denying that human cogni-
tion is merely computational for religious reasons.21 On the other
hand, some materialists support the argument, only to conclude, as
Penrose does, that human cognition can still be objectively evalu-
ated, but only with methods that go “beyond that limited scientif-
ic world-view that we hold today.”22 Thus we have both Christians
denying the argument, and non-Christians accepting the argument
only to draw materialistic conclusions. Barr, however—himself a
Catholic—accepts the Lucas-Penrose argument, and uses it to ad-
vance his program of effecting faith-science cohesion.
In terms of Christian philosophy, Barr’s defense of premise (1) has
the most obvious weaknesses. Barr contends that though the human
mind itself is not suitable for inspection, we only need to inspect a
computer program equivalent to it. Such a computer program, he
argues, would be able to inspect itself, for “one can certainly imagine
. . . a kind of machine whose internal parts and programming are
easily open to self-inspection.”23 Though one may readily grant the
phrase “open to inspection,” Barr’s phrase “open to self-inspection”
makes a much stronger claim. For the Christian, there is some-
thing philosophically dubious about a finite being—whether man or
machine—that is able to attain complete knowledge of itself.
In fact, the incompleteness theorems themselves almost seem to
preclude just such a possibility. In his attack against the Lucas–Pen-
rose argument, computer scientist Geoffrey LaForte writes, “One
conclusion to be drawn from the [incompleteness] results is that such
[exhaustive] self-knowledge can never be obtained, since the very
act of obtaining it changes one, as it were, into something new.”24
Here LaForte’s tone conveys a Christian sense of personal change,
even though, on the whole, he seems to support a materialist ap-
proach to human cognition. On this particular point, the incom-
pleteness theorems parallel the Christian notion that knowledge of
60 logos
God engenders transformation (e.g., 2 Cor 3:18). Thus, insofar as
the Penrose–Lucas argument requires a being (whether human or
artificial) to be capable of total self-comprehension (without chang-
ing into something new), the argument is unpersuasive to computer
scientists and Christians alike.
Premise (2), however, is also not as obviously agreeable to the
Christian as Barr might believe. Many opponents of the Lucas–
Penrose argument maintain that the human mind is an inconsistent
system, and this counterattack does not necessarily contradict the
Christian view of man. According to the Christian worldview, man
cannot be essentially inconsistent (i.e., capable of convincing himself
of anything in good conscience),25 and yet, unlike a machine, humans
are willing to entertain errors, to approximate truth, to jump to
conclusions, and then fill in the details later. In fact, there may be
something tremendously fruitful or creative about momentary ir-
rationality, as if it were the felix culpa of cognition. Computational
complexity theory seems to confirm that human reasoning does not
play by computational rules, for if it did, human decision-making
would be regularly thwarted by exponential-time computations just
as computers are. As neuroscientist Christopher Cherniak explained,
“Complexity theory provides a principled basis for raising the pos-
sibility that human beings [do not always execute] simple reasoning
tasks in ways that are guaranteed to be correct.”26
Furthermore, this sort of intrepid reasoning seems to permit the
very sort of abstraction that allows man to grasp the incompleteness
theorems themselves. By contrast, consider LaForte’s thought ex-
periment, which deals with how a theorem-computing robot might
wrestle with Gödel’s theorems:
[Link] Incompleteness
Christians have sometimes claimed that, by proving the existence
of unprovable truths, Gödel provided a mathematical justification
for belief in God’s existence. Gödel himself believed in God, but it
would be a mistake to interpret his incompleteness theorems as fur-
nishing mathematics with an absolutely unprovable (but nevertheless
true) proposition. Recall that the first of Gödel’s theorems imposes
mathematical incompleteness and divine ineffability 63
incompleteness only in a relative sense. Relative to a given theory
T, there are certain unprovable arithmetical truths, but a different
theory S may well be able to prove those same sentences. In other
words, all formal systems have their own unprovable truths, but no
truth is necessarily unprovable by all systems. In no way, therefore,
should the existence of relatively unprovable mathematical truths be
taken as evidence of the absolutely unprovable God.31
Thus, a much more nuanced approach is required if the incom-
pleteness theorems are to be brought into legitimate contact with
theology. Such an approach has been attempted at least once before:
in his chapter “Gödel, Thomas Aquinas, and the Unknowability of
God,” Catholic theologian Denys Turner took up the project of pro-
posing some tentative analogies.32 As a theologian,Turner’s approach
was to show how analogies of arithmetical incompleteness have long
been familiar to theological science. After reviewing and expanding
upon Turner’s analysis, I will make an attempt to invert his perspec-
tive by taking systematic theology as an object of formal logic.
Thomas Aquinas believed God’s existence to be rationally de-
monstrable, yet, by the success of this very demonstration, he be-
lieved that there must exist something demonstrable—God—that
cannot be entirely known. For, if God was univocally knowable, God
could not be God. Therefore, by making a rational demonstration of
God’s existence, Aquinas demonstrated the incompleteness of theol-
ogy. “It is on the grounds of what [theology] does know,” writes Turn-
er, “that it knows what it cannot know.”33 Both Gödel and Aquinas,
by different means, seem to have discovered a “rational procedure
of cracking open the closed circle of reason and exposing it to what
lies beyond.”34
This is about as much as Turner is willing to extend the Gödel–
Aquinas analogy. Here, however, I will push things a bit further. In
particular, it is notable that the immiscibility of completeness and
consistency seems to have been known, albeit informally, to Scho-
lastic theologians long before Gödel. Whenever philosophers, like
Aquinas, admit an unknowable God into their discourse, their theory
64 logos
of God cannot be complete if it is consistent (lest it be able to consis-
tently answer every question about God), and cannot be consistent
if it is complete (lest it assert that this completely knowable God
is the unknowable God). In mathematical logic, the incompleteness
theorems arise not because certain formal systems attempt to speak
about God, but because they are sufficiently complex to speak about
themselves. In either case, whether because of God or because of
self-referential definitions, a wild card has been introduced that forc-
es the rational system to concede the limits of its powers.
Certainly other intimations of computability theory could be
drawn from theology. Apophatic theology, for example, probably
has entertaining parallels with the logical concept of “semidecidabil-
ity.” Or one could explore Jason Wilson’s “natural theology incom-
pleteness” wherein “natural theologies require outside information/
revelation to determine if they are true.”35 But here I turn to the
project, not undertaken by Turner, of embedding the theological en-
terprise into a formal [Link], then, would the incompleteness
theorems have to say about theology?
Suppose the whole of divine revelation could be codified into a
formal system. Here, divine revelation could refer, for example, to
the assertions of the Scriptures, to a set of doctrinal declarations, to
a compendium of catechetical statements, or to a combination there-
of. Let T be the theory—called theology—that results from taking
divine revelation as its axioms. Assuming there are finitely many axi-
oms, or that these axioms form a denumerable set, the theory T is
“axiomatizable.”36
To apply Gödel’s first incompleteness theorem to T, it only re-
mains for T to be an extension of arithmetic.37 Are the axioms of
arithmetic deducible from divine revelation? This question is com-
plex: it depends on the scope of divine revelation and on the certain-
ty of arithmetical axioms.38 But the point of the arithmetic-extend-
ing condition in Gödel’s first incompleteness theorem is to ensure
that T is sufficiently well-accoutered so as to make reference to itself.
Since we have in mind an extremely powerful theory T, such a theory
mathematical incompleteness and divine ineffability 65
likely possesses the capacity for self-reference even without calling
upon the resources of arithmetic.39
These assumptions granted, I have a theory of theology T, which
is subject to the first incompleteness [Link] either T is incon-
sistent or it is incomplete. That is, either T deduces sentences that
are false, or else T fails to deduce everything that is true. Now, is T
consistent? If it is, as most Christians would grant, T is also subject to
the second incompleteness theorem.
Perhaps some believers assert the consistency of theology, T,
because they believe its consistency sentence CT to be divinely re-
vealed, that is, they believe that CT is included in T.40 Such reason-
ing, however, contradicts the second incompleteness theorem: T
cannot contain CT, otherwise T would be able to prove its own
consistency. Therefore, either T does not contain CT, or CT can-
not be expressed in terms of a classical language of mathematical
logic. I shall return to the second disjunct in the following section.
But if CT can indeed be expressed classically, then, regardless of
what it might seem, CT is not actually included in divine revelation.
Still, according to the second incompleteness theorem, the larger
theory T + CT is just as powerless to prove its own consistency, as
is the theory T + CT + C(T + CT), ad infinitum. Thus, in terms of clas-
sical logic, the consistency of divine revelation cannot be proven by
reference to divine revelation, even if divine revelation appears to
assert its own consistency. This claim is bound to disappoint those
who would wish to establish the inerrancy of the Scriptures by
reference to the Scriptures alone. But the upshot is that, in the
absence of some appropriate nonclassical logic, the consistency of
divine revelation seems to be a matter of pure and independent
faith: faith in divine revelation is not enough to cover faith in its
consistency. That is, if one wishes to assert the consistency of di-
vine revelation, one must believe something that is not divinely
revealed, nor formally derivable from divine revelation.41
Now I can return to the first incompleteness theorem: the Chris-
tian who takes the consistency of T by faith is forced to concede that
66 logos
T is incomplete. Of course, if T has the capacity to ask every possible
question about God, this conclusion is not surprising. For if such
a theology T was complete then it would be able to answer every
question about God. Then, as Turner observed, T would be inconsis-
tent, for this Being about which every question could be answered
could not be God. But this assumption—that T has the capacity to
ask e very possible question about God—is rather dubious. If inef-
fability can be predicated of God, then by definition he exceeds the
expressive capacity of all languages (and formal systems).Though we
might devise more and more powerful formal languages, there will
always remain unaskable questions about God.
Nevertheless, even if T cannot ask every possible question (e.g.,
about God), T can still ask unanswerable (theologically undecidable)
questions, according to the first incompleteness theorem. Further-
more, by the same theorem, not only certain questions about God
are undecidable within T, because undecidable questions arise even
in mundane subtheories like arithmetic. Thanks to Gödel, we see
that the set of unanswerable religious questions is utterly inexhaust-
ible, even in subdisciplines of theology that do not investigate the
nature of God.
In summary, under hypothetical conditions, divine revelation
could be recast into a formal system T subject to the incomplete-
ness theorems. Then, according to the first incompleteness theorem,
we would find the set of all theological truth to be far greater than
whatever T could deduce formally (assuming T’s consistency), not
only due to the unknowability of God, but as a consequence of the
arithmetic-exceeding capabilities intrinsic to theology. Next, ac-
cording to the second incompleteness theorem, the consistency of T
cannot be proved by T [Link], the Christian who wishes to assert
divine revelation’s consistency must make an additional act of faith,
an act on top of and independent of assent to divine revelation. Of
course, this act of faith need not be irrational: one might have good
reasons to believe divine revelation’s consistency, but these reasons
must draw from at least one premise external to divine revelation
mathematical incompleteness and divine ineffability 67
itself. Faith is apparently intrinsic to the theological enterprise, even
if divine revelation could be accepted without it.
Note, however, that our claims about theology apply just as well
to formal arithmetical systems. A consistent theory of divine revela-
tion can prove itself consistent no more than a consistent theory of
arithmetic (according to classical logic). But, if either is assumed to be
consistent (by faith, in the case of theology),42 then that theory will
be incomplete, that is, unable to prove all the truths its language can
express. Of course, faith and inscrutability have always been essen-
tial to the theological enterprise, for theology deals with God. Thus,
taken together, the incompleteness theorems merely seem to con-
firm from below what the ineffability of God requires from above:
faith in the logical possibility of God (i.e., theological consistency)
demands one to accept the inscrutability of his truths (i.e., theologi-
cal incompleteness).
Notes
1. The author is indebted to Dr. David Clemenson for his feedback on an early draft of
this article.
2. Here and elsewhere, Scripture is quoted from the New Revised Standard Version Catholic
Edition.
3. Marie Liguori Ewald, trans., “Homily on Psalm 115 (116B),” in The Homilies of Saint
Jerome (Washington, DC: The Catholic University of America Press, 1964), 1: 294.
Strictly speaking, Jerome was incorrect: the Epimenides paradox is soluble as fol-
lows. Suppose David knew that there was one man who always told the truth—Jesus
Christ, for example. Then by saying “Every man is a liar,” David would be lying, and
no contradiction would be generated. He would be wrong but not paradoxical.
4. This statement is subject to two caveats, which we express coarsely. First, the limita-
tions prescribed by the incompleteness theorems only apply to formal systems with
sufficient linguistic capacity for arithmetical self-reference. Second, such formal sys-
tems are based on particular codifications of logical reasoning. These codifications are
thought to be the only ones capable of formalizing mathematics, an assumption to
which we return in the concluding section.
5. See Russell W. Howell, “The Matter of Mathematics,” Perspectives on Science and Chris-
tian Faith 67, no. 2 (2015): 77.
6. Torkel Franzén, Gödel’s Theorem: An Incomplete Guide to Its Use and Abuse (Wellesley,
MA: A K Peters, 2005). See also Torkel Franzén, “The Popular Impact of Gödel’s In-
completeness Theorem,” Notices of the American Mathematical Society 53, no. 4 (2006):
440–43.
7. If we should like to encode not only sentences, but terms, our simplistic example
runs into a problem. For example, the term 74,998 receives the same code number
as the sentence 1 = 3. This could be avoided, however, by letting the numeral n
instead have the code number 2n + 1, since the code numbers of our sentences are
always even.
8. The sentence G is made to assert its own nonprovability in the standard interpreta-
tion of the formal system’s language, but not necessarily in other interpretations.
Unless otherwise specified, semantic concepts such as truth and completeness will
refer to this standard interpretation.
9. Caution is required here. G is defined with respect to a formal system. We could cer-
tainly prove the G-sentence of one formal system using the apparatus of a different
arithmetical formal system, but that second system, if consistent, would have its own
unprovable G-sentence.
10. Non-classical “paraconsistent logics” (formal systems which tolerate some degree of
inconsistency) have been devised for which this result does not hold.
11. To be precise, “theory of arithmetic” refers to an extension of Peano arithmetic. A
theory T that can “prove its own consistency” is one for which the consistency sen-
tence for T is provable in T. For details, see George S. Boolos, John P. Burgess, and
72 logos
Richard C. Jeffrey, Computability and Logic, 5th ed. (New York, NY: Cambridge Uni-
versity Press, 2010), 232–33.
12. Again, a “formal system” refers to a system that has (1) enough linguistic power to
capture the essential elements of arithmetic, and that is (2) based upon a standard
codification of logic. See note 3. Of course, this summary statement holds true vacu-
oulsy if in fact, no consistent system of arithmetic even exists.
13. John D. Barrow, “Gödel and Physics,” in Kurt Gödel and the Foundations of Mathematics,
ed. Matthias Baaz, Christos H. Papadimitriou, Hilary W. Putnam, Dana S. Scott, and
Charles L. Harper Jr. (New York, NY: Cambridge University Press, 2011), 259. See
also John D. Barrow, “Mathematical Jujitsu: Some Informal Thoughts About Gödel
and Physics,” Complexity 5, no. 5 (2000): 28–34.
14. Stanley Jaki, The Relevance of Physics (Chicago: University of Chicago Press, 1966),
129.
15. Barrow, “Gödel and Physics,” 260.
16. Stephen Hawking, “Gödel and the End of Physics,” 2002, [Link]
.uk/[Link].
17. Barrow, “Gödel and Physics,” 265.
18. To be precise, the incompleteness theorems do not necessarily arise out of the
Riemannian geometry upon which general relativity is based, according to Barrow,
“Gödel and Physics,” 265.
19. Adding this Gödel sentence to the program’s axioms does not resolve the paradox,
for the program will then have a new Gödel sentence that it again cannot assert,
though its human counterpart still can.
20. Stephen M. Barr, Modern Physics and Ancient Faith, (Notre Dame, IN: University of
Notre Dame Press, 2003), 213–19.
21. Howell, for example, does not favor the Lucas–Penrose argument, although his dis-
cussion of it is sparing (Russell W. Howell, “The Matter of Mathematics,” Perspectives
on Science and Christian Faith 67, no. 2 [2015] 77). We hope to make some of his rea-
soning explicit in what follows. Another Christian dissenting from Lucas–Penrose is
logician Michael Detlefson. See Michael Detlefsen, “Comment on Roger Penrose’s
‘Mathematics, the Mind, and the Physical World’: Gödel’s Theorems and Platonism,”
in Meaning in Mathematics, ed. John Polkinghorne (Oxford: Oxford University Press,
2011), 46–47.
22. Roger Penrose, “On Understanding Understanding,” International Studies in the Philoso-
phy of Science 11, no. 1 (1997): 19. In the life sciences, for instance, Newtonian models
of biological systems are increasingly seen as inadequate, and yet hope is held out that
nature-inspired computation or quantum computing could effectively simulate such
systems (Plamen L. Simeonov, “Integral Biomathics: A Post-Newtonian View Into the
Logos of Bios,” Progress in Biophysics and Molecular Biology 102 [2010]: 85–121).
23. Barr, Modern Physics, 216.
24. Geoffrey LaForte, Patrick J. Hayes, and Kenneth M. Ford. “Why Gödel’s Theorem
Cannot Refute Computationalism,” Artificial Intelligence 103 (1998): 275.
mathematical incompleteness and divine ineffability 73
25. See, for example, Rom 1:20–21.
26. Christopher Cherniak, Minimal Rationality, (Cambridge, MA: MIT Press, 1986), 81.
27. LaForte, “Gödel’s Theorem,” 275.
28. Note well: it is unclear that real computers (as distinct from abstract Turing ma-
chines) are capable of any unlimited operation.
29. Second Vatican Council, Pastoral Constitution on the Church in the Modern World
Gaudium et Spes (December 7, 1965), §15. Emphasis added.
30. Gregory of Nyssa, Making of Man 2 (cited in Robert Louis Wilken, The Spirit of Early
Christian Thought [New Haven, CT:Yale University Press, 2003], 153).
31. Assuming, of course, that God’s existence cannot be deduced a priori. Here, as else-
where, provability refers to a syntactic deduction procedure, not informal inference
nor inductive reasoning.
32. Denys A. Turner, “Gödel, Thomas Aquinas, and the Unknowability of God,” in Kurt
Gödel and the Foundations of Mathematics, ed. Matthias Baaz, Christos H. Papadimitriou,
Hilary W. Putnam, Dana S. Scott, and Charles L. Harper Jr. (New York: Cambridge
University Press, 2011), 277–96.
33. Turner, “Gödel, Thomas Aquinas,” 294. While much of contemporary theology re-
jects Aquinas’s proof of God’s existence, Turner observes that Gödel’s results are
more consonant “with some premodern theologies than [they would be] with the
theological consensus of our times” (Turner, “Gödel, Thomas Aquinas,” 279).
34. Turner, “Gödel, Thomas Aquinas,” 295.
35. Jason Wilson, “Integration of Faith and Mathematics from the Perspectives of Truth,
Beauty, and Goodness,” Perspectives on Science and Christian Faith 67, no. 2 (2015):
102–103.
36. Formally, the set of axioms must be recursive in order to be axiomatizable.
37. Here and elsewhere, by arithmetic, we mean a theory of arithmetic sufficiently
strong to generate the incompleteness theorems.
38. This issue should not be quickly dismissed. If one claims that divine revelation is
consistent and is able to deduce the axioms of arithmetic, one must be ready to claim
that divine revelation is a guarantor for the consistency of inductive arithmetic. In-
deed, there are at least some reasons to doubt the consistency of inductive (Peano)
arithmetic: Edward Nelson, “Warning Signs of a Possible Collapse of Contempo-
rary Mathematics,” in Infinity: New Research Frontiers, ed. Michael Heller and W. Hugh
Woodin (Cambridge, UK: Cambridge University Press, 2011), 76–88.
39. Recall, however, Barrow’s contention in the case of physics: there are some very
elegant formal systems that do not generate incompleteness phenomena. Still, I will
proceed for the sake of argument.
40. For example, Catholics might construe the doctrinal statement “Truth cannot con-
tradict truth” (Pope Leo XIII, Providentissimus Deus, §23) as a guarantee that Church
doctrine does not contradict itself. Similarly, 2 Tim 3:16 (“All scripture is inspired by
God”) seems to have Scripture asserting its own consistency.
41. The Catholic, on this point, will be interested in the following claim from Donum
74 logos
Veritatis, §23: “When the Magisterium proposes ‘in a definitive way’ truths concern-
ing faith and morals, which, even if not divinely revealed, are nevertheless strictly
and intimately connected with Revelation, these must be firmly accepted and held.”
42. In mathematics, the consistency of a theory T can be proved by some other theory T 1,
but the consistency of that other theory is subject to the same incompleteness theo-
rems. Thus, within the framework of classical logic, it is impossible to definitively
demonstrate a given theory’s consistency, even though that theory’s consistency may
be granted for other (nonformal) reasons.
43. To be clear, we refer here to the consistency of ZFC, as opposed to the consistency of
inductive [Link] proofs of the incompleteness theorems can be mechanically
verified using the axioms of ZFC (see Natarajan Shankar, Metamathematics, Machines
and Gödel’s Proof [Cambridge, UK: Cambridge University Press, 1994]). But if ZFC is
inconsistent, then Gödel’s theorems would lose their force, and modern mathemat-
ics would face a far graver dilemma than that posed by incompleteness. Conversely, if
the incompleteness theorems are presumed valid, then an absolutely definitive dem-
onstration of ZFC’s consistency seems unattainable (see note 41). Thus we have a
kind of vicious circle of uncertainty: the incompleteness theorems ultimately impugn
their own veracity.
44. See, for example, Joseph W. Dauben, “Georg Cantor and Pope Leo XIII: Mathemat-
ics, Theology, and the Infinite,” Journal of the History of Ideas (1977): 85–108.
45. Solomon Feferman, In the Light of Logic (New York: Oxford University Press, 1998).