Putnam - Models and Reality
Putnam - Models and Reality
Putnam - Models and Reality
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Association for Symbolic Logic is collaborating with JSTOR to digitize, preserve and extend
access to The Journal of Symbolic Logic
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
THE JOURNAL OF SYMBOLIC LoGic
Volume 45, Number 3, Sept. 1980
HILARY PUTNAM
464
? 1980, Association for Symbolic Logic
0022-48 12/80/4503-0004/$05.75
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 465
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
466 HILARY PUTNAM
Now the argument that Skolem gave, and that shows that "the intuitive notion
of a set" (if there is such a thing) is not "captured" by any formal system, shows
that even a formalization of total science (if one could construct such a thing), or
even a formalization of all our beliefs (whether they count as "science" or not),
could not rule out denumerable interpretations, and, afortiori, such a formalization
could not rule out unintended interpretations of this notion.
This shows that "theoretical constraints", whether they come from set theory
itself or from "total science", cannot fix the interpretation of the notion set in the
"intended" way. What of "operational constraints"?
Even if we allow that there might be a denumerable infinity of measurable "mag-
nitudes", and that each of them might be measured to arbitrary rational accuracy
(which certainly seems a utopian assumption), it would not help. For, by the
"downward LUwenheim-Skolem Theorem", we can find a countable submodel
of the "standard" model (if there is such a thing) in which countably many pre-
dicates (each of which may have countably many things in its extension) have their
extensions preserved. In particular, we can fix the values of countable many
magnitudes at all rational space-time points, and still find a countable submodel
which meets all the constraints. In short, there certainly seems to be a countable
model of our entire body of belief which meets all operational constraints.
The philosophical problem appears at just this point. If we are told, "axiomatic
set theory does not capture the intuitive notion of a set", then it is natural to think
that something else-our "understanding"-does capture it. But what can our
"understanding" come to, at least for a naturalistically minded philosopher, which
is more than the way we use our language? The Skolem argument can be extended,
as we have just seen, to show that the total use of the language (operational plus
theoretical constraints) does not "fix" a unique "intended interpretation" any more
than axiomatic set theory by itself does.
This observation can push a philosopher of mathematics in two different ways.
If he is inclined to Platonism, he will take this as evidence that the mind has mys-
terious faculties of "grasping concepts" (or "perceiving mathematical objects")
which the naturalistically minded philosopher will never succeed in giving an
account of. But if he is inclined to some species of verificationism (i.e., to indenti-
fying truth with verifiability, rather than with some classical "correspondence
with reality") he will say, "Nonsense! All the 'paradox' shows is that our under-
standing of 'The real numbers are nondenumerable' consists in our knowing
what it is for this to be proved, and not in our 'grasp' of a 'model'." In short, the
extreme positions-Platonism and verificationism-seem to receive comfort from
the Ldwenheim-Skolem Parodox; it is only the "moderate" position (which tries
to avoid mysterious "perceptions" of "mathematical objects" while retaining a
classical notion of truth) which is in deep trouble.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 467
model" if they all satisfy the same sentences? What we want to know as mathema-
ticians is what sentences of set theory are true; we do not want to have the sets
themselves in our hands.
Unfortunately, the argument can be extended. First of all, the theoretical con-
straints we have been speaking of must, on a naturalistic view, come from only
two sources: they must come from something like human decision or convention,
whatever the source of the "naturalness" of the decisions or conventions may be,
or from human experience, both experience with nature (which is undoubtedly the
source of our most basic "mathematical intuitions", even if it be unfashionable
to say so), and experience with "doing mathematics". It is hard to believe that
either or both of these sources together can ever give us a complete set of axioms
for set theory (since, for one thing, a complete set of axioms would have to be
nonrecursive, and it is hard to envisage coming to have a nonrecursive set of axioms
in the literature or in our heads even in the unlikely event that the human race
went on forever doing set theory); and if a complete set of axioms is impossible,
and the intended models (in the plural) are singled out only by theoretical plus
operational constraints then sentences which are independent of the axioms which
we will arrive at in the limit of set-theoretic inquiry really have no determinate
truth value; they are just true in some intended models and false in others.
To show what bearing this fact may have on actual set-theoretic inquiry, I will
have to digress for a moment into technical logic. In 1938 Godel put forward a
new axiom for set theory: the axiom " V = L". Here L is the class of all construc-
tible sets, that is, the class of all sets which can be defined by a certain constructive
procedure if we pretend to have names available for all the ordinals, however large.
(Of course, this sense of "constructible" would be anathema to constructive
mathematicians.) V is the universe of all sets. So " V = L" just says all sets are
constructible. By considering the inner model for set theory in which "V = L" is
true, G6del was able to prove the relative consistency of ZF and ZF plus the axiom
of choice and the generalized continuum hypothesis.
" V = L" is certainly an important sentence, mathematically speaking. Is it true?
G6del briefly considered proposing that we add " V = L" to the accepted axioms
for set theory, as a sort of meaning stipulation, but he soon changed his mind. His
later view was that "V = L" is really false, even though it is consistent with set
theory, if set theory is itself consistent.
G6del's intuition is widely shared among working set theorists. But does this
"intuition" make sense?
Let MAG be a countable set of physical magnitudes which includes all magni-
tudes that sentient beings in this physical universe can actually measure (it certainly
seems plausible that we cannot hope to measure more than a countable number of
physical magnitudes). Let OP be the "correct" assignment of values; that is, the as-
signment which assigns to each member of MAG the value that that magnitude
actually has at each rational space-time point. Then all the information "opera-
tional constraints" might give us (and, in fact, infinitely more) is coded into OP.
One technical term: an co-model for a set theory is a model in which the natural
numbers are ordered as they are "supposed to be"; that is, the sequence of "natural
numbers" of the model is an w-sequence.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
468 HILARY PUTNAM
2 Barwise has proved the much stronger theorem that every countable model of ZF has a proper
end extension which is a model of ZF + V = L (in Infinitary methods in the model theory of set
theory, published in Logic Colloquium '69). The theorem in the text was proved by me before 1963.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 469
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
470 HILARY PUTNAM
'This is a very counterintuitive consequence of the axiom of choice. Call two objects A, B
"congruent by finite decomposition" if they can be divided into finitely many disjoint point
sets Al, ..., An, B1, ...By such that A = A, U A2U ... UAn, B = B1 UB2U ... U B, and (for i =
1, 2, ..., n) Ai is congruent to Bi. Then Tarski and Banach showed that all spheres are congruent
by finite decomposition.
I This axiom, first studied by J. Mycielski (On the axiom of determinacy", Fundamenta Ma-
thematicae, 1963) asserts that infinite games with perfect information are determined, i.e. there is a
winning strategy for either the first or second player. AD (the axiom of determinacy) implies the
existence of a nontrivial countably additive two-valued measure on the real numbers, contradict-
ing a well-known consequence of the axiom of choice.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 471
Is the problem a problem with the notion of a "set"? It would be natural to suppose
that the problem Skolem points out, the problem of a surprising "relativity" of our
notions, has to do with the notion of a "set", given the various problems which are
known to surround that notion, or, at least, has to do with the problem of reference
to "mathematical objects". But this is not so.
To see why it is not so, let us consider briefly the vexed problem of reference
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
472 HILARY PUTNAM
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 473
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
474 HILARY PUTNAM
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 475
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
476 HILARY PUTNAM
various models. If we further agree with Wittgenstein that the notion of truth re-
quires a public language (or requires at least states of the self at more than one time
-that a "private language for one specious present" makes no sense), then even
my present sense data are in this same boat . . .. In short, one can "Skolemize"
absolutely everything. It seems to be absolutely impossible to fix a determinate
reference (without appeal to nonnatural mental powers) for any term at all. If we
apply the argument to the very metalanguage we use to talk about the predica-
ment . . . ?
The same problem has even surfaced recently in the field of cognitive psychology.
The standard model for the brain/mind in this field is the modern computing
machine. This computing machine is thought of as having something analogous to
a formalized language in which it computes. (This hypothetical brain language has
even received a name-"mentalese".) What makes the model of cognitive psychology
a cognitive model is that "mentalese" is thought to be a medium whereby the brain
constructs an internal representation of the external world. This idea runs imme-
diately into the following problem: if "mentalese" is to be a vehicle for describing
the external world, then the various predicate letters must have extensions which
are sets of external things (or sets of n-tuples of external things). But if the way
"mentalese" is "understood" by the deep structures in the brain that compute,
record, etc. in this "language" is via what artificial intelligence people call "pro-
cedural semantics"-that is, if the brain's program for using "mentalese" comprises
its entire "understanding" of "mentalese"-where the program for using "men-
talese", like any program, refers only to what is inside the computer- then how do
extensions ever come into the picture at all? In the terminology I have been employ-
ing in this address, the problem is this: if the extension of predicates in "mentalese"
is fixed by the theoretical and operational constraints "hard wired in" to the brain,
or even by theoretical and operational constraints that it evolves in the course of
inquiry, then these will not fix a determinate extension for any predicate. If thinking
is ultimately done in "mentalese", then no concept we have will have a determinate
extension. Or so it seems.
The bearing of causal theories of reference. The term "causal theory of reference"
was originally applied to my theory of the reference of natural kind terms and
Kripke's theory of the reference of proper names. These theories did not attempt
to define reference, but rather attempted to say something about how reference is
fixed, if it is not fixed by associating definite descriptions with the terms and names
in question. Kripke and I argued that the intention to preserve reference through a
historical chain of uses and the intention to cooperate socially in the fixing of
reference make it possible to use terms successfully to refer although no one definite
description is associated with any term by all speakers who use that term. These
theories assume that individuals can be singled out for the purpose of a "naming
ceremony" and that inferences to the existence of definite theoretical entities (to
which names can then be attached) can be successfully made. Thus these theories
did not address the question as to how any term can acquire a determinate reference
(or any gesture, e.g., pointing-of course, the "reference" of gestures is just as
problematic as the reference of terms, if not more so). Recently, however, it has
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 477
been suggested by various authors that some account can be given of how at least
some basic sorts of terms refer in terms of the notion of a "causal chain". In one
version,5 a version strikingly reminiscent of the theories of Ockham and other 14th
century logicians, it is held that a term refers to "the dominant source" of the beliefs
that contain the term. Assuming we can circumvent the problem that the dominant
cause of our beliefs concerning electrons may well be textbooks,6 it is important to
notice that even if a correct view of this kind can be elaborated, it will do nothing
to resolve the problem we have been discussing.
The problem is that adding to our hypothetical formalized language of science a
body of theory titled "causal theory of reference" is just adding more theory. But
Skolem's argument, and our extensions of it, are not affected by enlarging the
theory. Indeed, you can even take the theory to consist of all true sentences, and
there will be many models-models differing on the extension of every term not
fixed by OP (or whatever you take OP to be in a given context)-which satisfy
the entire theory. If "refers" can be defined in terms of some causal predicate or
predicates in the metalanguage of our theory, then, since each model of the object
language extends in an obvious way to a corresponding model of the metalanguage,
it will turn out that, in each model M, references is definable in terms of causesM;
but, unless the word 'causes' (or whatever the causal predicate or predicates may
be) is already glued to one definite relation with metaphysical glue, this does not
fix a determinate extension for 'refers' at all.
This is not to say that the construction of such a theory would be worthless as
philosophy or as natural science. The program of cognitive psychology already al-
luded to-the program of describing our brains as computers which construct an
"internal representation of the environment" seems to require that "mentalese"
utterances be, in some cases at least, describable as the causal product of devices in
the brain and nervous system which "transduce" information from the environ-
ment, and such a description might well be what the causal theorists are looking for.
The program of realism in the philosophy of science-of empirical realism, not
metaphysical realism-is to show that scientific theories can be regarded as better
and better representations of an objective world with which we are interacting; if
such a view is to be part of science itself, as empirical realists contend it should be,
then the interactions with the world by means of which this representation is formed
and modified must themselves be part of the subject matter of the representation.
But the problem as to how the whole representation, including the empirical theory
of knowledge that is a part of it, can determinately refer is not a problem that can
be solved by developing more and better empirical theory.
Ideal theories and truth. One reaction to the problem I have posed would be to
say: there are many ideal theories in the sense of theories which satisfy the opera-
tional constraints, and in addition have all the virtues (simplicity, coherence, con-
5 Cf. Gareth Evans, The causal theory of names, Aristotelian Society Supplementary Volume
XLVII, pp. 187-208, reprinted in Naming, necessity and natural kinds, (Stephen P. Schwartz,
Editor), Cornell University Press, 1977.
6 Evans handles this case by saying that there are appropriateness conditions on the type of cau
chain which must exist between the item referred to and the speaker's body of information.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
478 HILARY PUTNAM
training the axiom of choice, whatever) that humans like to demand. But there are
no "facts of the matter" not reflected in constraints on ideal theories in this sense.
Therefore, what is really true is what is common to all such ideal theories; what is
really false is what they all deny; all other statements are neither true nor false.
Such a reaction would lead to too few truths, however. It may well be that there
are rational beings-even rational human species-which do not employ our color
predicates, or who do not employ the predicate "person", or who do not employ
the predicate "earthquake".7 I see no reason to conclude from this that our talk of
red things, or of persons, or of earthquakes, lacks truth value. If there are many
ideal theories (and if "ideal" is itself a somewhat interest-relative notion), if there
are many theories which (given appropriate circumstances) it is perfectly rational
to accept, then it seems better to say that, insofar as these theories say different (and
sometimes, apparently incompatible) things, that some facts are "soft" in the sense
of depending for their truth value on the speaker, the circumstances of utterance,
etc. This is what we have to say in any case about cases of ordinary vagueness,
about ordinary causal talk, etc. It is what we say about apparently incompatible
statements of simultaneity in the special theory of relativity. To grant that there is
more than one true version of reality is not to deny that some versions are false.
It may be, of course, that there are some truths that any species of rational in-
quirers would eventually acknowledge. (On the other hand, the set of these may be
empty, or almost empty.) But to say that by definition these are all the truths there
are is to redefine the notion in a highly restrictive way. (It also assumes that the
notion of an "ideal theory" is perfectly clear; an assumption which seems plainly
false.)
Intuitionism. It is a striking fact that this entire problem does not arise for the
standpoint of mathematical intuitionism. This would not be a surprise to Skolem:
it was precisely his conclusion that "most mathematicians want mathematics to
deal, ultimately, with performable computing operations and not to consist of
formal propositions about objects called this or that."
In intuitionism, knowing the meaning of a sentence or predicate consists in asso-
ciating the sentence or predicate with a procedure which enables one to recognize
when one has a proof that the sentence is constructively true (i.e., that it is possible
to carry out the constructions that the sentence asserts can be carried out) or that
the predicate applies to a certain entity (i.e., that a certain full sentence of the
predicate is constructively true). The most striking thing about this standpoint is
that the classical notion of truth is nowhere used-the semantics is entirely given in
terms of the notion of "constructive proof", including the semantics of "constructive
proof" itself.
Of course, the intuitionists do not think that "constructive proof' can be for-
malized, or that "mental constructions" can be identified with operations in our
brains. Generally, they assume a strongly intentionalist and a prioristic posture in
philosophy-that is, they assume the existence of mental entities called "meanings"
and of a special faculty of intuiting constructive relations between these entities.
I For a discussion of this very point, cf. David Wiggins, Truth, invention and the meaning of
life, British Academy, 1978.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 479
These are not the aspects of intuitionism I shall be concerned with. Rather I wish
to look on intuitionism as an example of what Michael Dummett has called "non-
realist semantics"-that is, a semantic theory which holds that a language is com-
pletely understood when a verification procedure is suitably mastered, and not
truth conditions (in the classical sense) are learned.
The problem with realist semantics-truth-conditional semantics-as Dummett
has emphasized, is that if we hold that the understanding of the sentences of, say,
set theory consists in our knowledge of their "truth conditions", then how can we
possibly say what that knowledge in turn consists in? (It cannot, as we have just
seen, consist in the use of language or "mentalese" under the control of operational
plus theoretical constraints, be they fixed or evolving, since such constraints are
too weak to provide a determinate extension for the terms, and it is this that the
realist wants.)
If, however, the understanding of the sentences of a mathematical theory consists
in the mastery of verification procedures (which need not be fixed once and for
all-we can allow a certain amount of "creativity"), then a mathematical theory
can be completely understood, and this understanding does not presuppose the
notion of a "model" at all, let alone an "intended model".
Nor does the intuitionist (or, more generally, the "nonrealist" semanticist) have
to foreswearforever the notion of a model. He has to foreswear reference to models
in his account of understanding; but, once he has succeeded in understanding a rich
enough language to serve as a metalanguage for some theory T (which may itself be
simply a sublanguage of the metalanguage, in the familiar way), he can define 'true
in T' a la Tarski, he can talk about "models" for T, etc. He can even define 'refer-
ence' or ('satisfaction') exactly as Tarski did.
Does the whole "Skolem Paradox" arise again to plague him at this stage? The
answer is that it does not. To see why it does not, one has to realize what the
"existence of a model" means in constructive mathematics.
"Objects" in constructive mathematics are given through descriptions. Those
descriptions do not have to be mysteriously attached to those objects by some
nonnatural process (or by metaphysical glue). Rather the possibility of proving
that a certain construction (the "sense", so to speak, of the description of the
model) has certain constructive properties is what is asserted and all that is asserted
by saying the model "exists". In short, reference is given through sense, and sense is
given through verification-procedures and not through truth-conditions. The "gap"
between our theory and the "objects" simply disappears-or, rather, it never ap-
pears in the first place.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
480 HILARY PUTNAM
8 To the suggestion that we identify truth with being verified, or accepted, or accepted in the
run, it may be objected that a person could reasonably, and possibly truly, make the assertion:
A; but it could have been the case that A and our scientific development differ in such a
way to make A part of the ideal theory accepted in the long run; in that circumstance, it
would have been the case that A but it was not true that A.
This argument is fallacious, however, because the different "scientific development" means here
the choice of a different version; we cannot assume the sentence -A- has a fixed meaning independ-
ent of what version we accept.
More deeply, as Michael Dummett first pointed out, what is involved is not that we identify
truth with acceptability in the long run (is there a fact of the matter about what would be accepted
in the long run?), but that we distinguish two truth-related notions: the internal notion of truth
("snow is white" is true if and only if snow is white), which can be introduced into any theory at
all, but which does not explain how the theory is understood (because "snow is white" is true is
understood as meaning that snow is white and not vice versa, and the notion of verification, no longer
thought of as a mere index of some theory-independent kind of truth, but as the very thing in terms
of which we understand the language.
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
MODELS AND REALITY 481
realist semantics; it is simply prior to it, in the sense that it is the "nonrealist"
semantics that must be internalized if the language is to be understood.
Even if it is not inconsistent with realist semantics, taking the nonrealist seman-
tics as our picture of how the language is understood undoubtedly will affect the
way we view questions about reality and truth. For one thing, verification in em-
pirical science (and, to a lesser extent, in mathematics as well, perhaps) sometimes
depends on what we before called "decision" or "convention". Thus facts may, on
this picture, depend on our interests, saliencies and decisions. There will be many
"soft facts". (Perhaps whether V = L or not is a "soft fact".) I cannot, myself,
regret this. If appearance and reality end up being endpoints on a continuum rather
than being the two halves of a monster Dedekind cut in all we conceive and do not
conceive, it seems to me that philosophy will be much better off. The search for the
"furniture of the Universe" will have ended with the discovery that the Universe is
not a furnished room.
Where did we go wrong?-The problem solved. What Skolem really pointed out
is this: no interesting theory (in the sense of first-order theory) can, in and of itself,
determine its own objects up to isomorphism. Skolem's argument can be extended
as we saw, to show that if theoretical constraints do not determine reference, then
the addition of operational constraints will not do it either. It is at this point that
reference itself begins to seem "occult"; that it begins to seem that one cannot be
any kind of a realist without being a believer in nonnatural mental powers. Many
moves have been made in response to this predicament, as we noted above. Some
have proposed that second-order formalizations are the solution, at least for mathe-
matics; but the "intended" interpretation of the second-order formalism is not
fixed by the use of the formalism (the formalism itself admits so-called "Henkin
models", i.e., models in which the second-order variables fail to range over the full
power set of the universe of individuals), and it becomes necessary to attribute to
the mind special powers of "grasping second-order notions". Some have proposed
to accept the conclusion that mathematical language is only partially interpreted,
and likewise for the language we use to speak of "theoretical entities" in empirical
science; but then are "ordinary material objects" any better off? Are sense data
better off? Both Platonism and phenomenalism have run rampant at different
times and in different places in response to this predicament.
The problem, however, lies with the predicament itself. The predicament only
is a predicament because we did two things: first, we gave an account of under-
standing the language in terms of programs and procedures for using the language
(what else?); then, secondly, we asked what the possible "models" for the language
were, thinking of the models as existing "out there" independent of any description.
At this point, something really weird had already happened, had we stopped to
notice. On any view, the understanding of the language must determine the
reference of the terms, or, rather, must determine the reference given the context o
use. If the use, even in a fixed context, does not determine reference, then use is not
understanding. The language, on the perspective we talked ourselves into, has a
full program of use; but it still lacks an interpretation.
This is the fatal step. To adopt a theory of meaning according to which a lan-
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms
482 HILARY PUTNAM
guage whose whole use is specified still lacks something-viz. its "interpretation"-
is to accept a problem which can only have crazy solutions. To speak as if this were
my problem, "I know how to use my language, but, now, how shall I single out an
interpretation?" is to speak nonsense. Either the use already fixes the "interpreta-
tion" or nothing can.
Nor do "causal theories of reference", etc., help. Basically, trying to get out of
this predicament by these means is hoping that the world will pick one definite
extension for each of our terms even if we cannot. But the world does not pick
models or interpret languages. We interpret our languages or nothing does.
We need, therefore, a standpoint which links use and reference in just the way
that the metaphysical realist standpoint refuses to do. The standpoint of "non-
realist semantics" is precisely that standpoint. From that standpoint, it is trivial
to say that a model in which, as it might be, the set of cats and the set of dogs are
permuted (i.e., 'cat' is assigned the set of dogs as its extension, and 'dog' is as-
signed the set of cats) is "unintended" even if corresponding adjustments in the
extensions of all the other predicates make it end up that the operational and the-
oretical constraints of total science or total belief are all "preserved". Such a model
would be unintended because we do not intend the word 'cat' to refer to dogs. From
the metaphysical realist standpoint, this answer does not work; it just pushes the
question back to the metalanguage. The axiom of the metalanguage, " 'cat' refers
to cats" cannot rule out such an unintended interpretation of the object language,
unless the metalanguage itself already has had its intended interpretation singled
out; but we are in the same predicament with respect to the metalanguage that we
are in with respect to the object language, from that standpoint, so all is in vain.
However, from the viewpoint of "nonrealist" semantics, the metalanguage is
completely understood, and so is the object language. So we can say and understand,
" 'cat' refers to cats". Even though the model referred to satisfies the theory, etc.,
it is "unintended"; we recognize that it is unintended from the description through
which it is given (as in the intuitionist case). Models are not lost noumenal waifs
looking for someone to name them; they are constructions within our theory itself,
and they have names from birth.
DEPARTMENT OF PHILOSOPHY
HARVARD UNIVERSITY
This content downloaded from 160.39.16.248 on Wed, 21 Aug 2019 04:23:58 UTC
All use subject to https://about.jstor.org/terms