W.V.

Quine: Perspectives on Logic,

Science and Philosophy
- -- - --- -- -- -- - -

Interview by Bradley Edmister and Michael OJShea

HRP: What is the role of philosophy vis-a-vis science?

Quine: I think of philosophy as continuous with science, but philosophy differs
by degree in various respects. Philosophy undertakes to analyze the general,
basic concepts of science - the sort of concepts the practicing scientist will typ-
ically take for granted. These are such basic notions as truth, existence, and
necessity. Also, philosophy investigates questions of evidence for science -
that's epistemology. It seeks a better understanding of the tremendous trans-
formation that takes place between the input that we receive through the irrita-
tion of our sensory surfaces, and our torrential output in the form of scientific
theory. It tries to analyze theory, and see how much of it is really dictated by
the input ("by nature," we say, but that's only going to be in the input), and
how much is only a matter of our accommodation and organization of it.
These are considerations that aren't ordinarily taken up by any particular sci-
ence.
In these studies, philosophy will sometimes elicit paradoxes, which the sci-
entist, even if he is told about them, isn't likely to worry about. In normal sci-
entific practice, he can simply dodge that
end of his theory. But the philosopher is
going to be concerned.

HRP: So the existence of paradoxes in

philosophy of science doesn't affect the
workings of scientific theory?
Quine: Right. This is brought up most
dramatically in the familiar paradoxes of W. K Quine is Edg-ar Pierce
Professor of Philosophy,
Emeritus, at Harvard
set theory - Russell's Paradox and the
like. Even mathematicians, as a whole, University. Among his many
didn't worry about them, because they books and articles are Web of
weren't going t o be dealing with self- Belief, Pursuit of Truth, Word
membership in classes, o r classes of all and Object, Roots of Reference
non-self-membered classes. They worked and Set Theory and Its Logic.
in mathematical domains where, when
intuitively surveying the assumptions and
axioms involved, they felt they were on
solid ground. It's the sort of thing that
falls quite naturally into the philosopher's
domain.
HRP: That view seems to reduce the work of philosophy to simply tying
up the loose ends of science. Is that accurate?
Quine: Yes, I think "tying up loose ends" is a good way of condensing philoso-
phy's purpose.

HRP: How much science should a philosopher know in order to do his

job competently?
Quine: It's important that he have a respectable grasp, at least at the under-
graduate level, of a hard science. It's extremely important to have had the
experience of really knowing something without a shadow of doubt, even if it
takes complicated argument, and having what one feels is a firm basis of evi-
dence. Ideal for this purpose is mathematics, especially mathematical logic.
I have always felt that a good course in modern logic ought to be required,
not only of every candidate for a philosophy degree, but for every undergradu-
ate in every field. It could even be valuable for students specializing in the lit-
erary domain, where one doesn't aspire to logic's sort of certainty and convic-
tion. It's important for those students to see the difference.

HRP: What is the role of first-order logic in the scientific project, in

physics, for example?
Quine: In relation to physics, I don't see much of a role. I think that as far as
elementary logic is concerned, the common sense of the physicist today (and of
centuries ago, perhaps, before modern logic began) is enough. But modern
logic is important for a systematic understanding of theoretical relationships;
and when it comes to more analytical studies, both in foundations of mathe-
matics and in philosophical analysis of concepts generally, logic is vital.
The illumination that modern logic has brought to philosophy of mathe-
matics is the most illuminating instance. Godel's incompleteness proof contra-
dicted what everybody, even mathematicians, had taken for granted, namely,
that mathematical truth consists in demonstrability. [It was thought that] you
may not find the proof, but the proof can be found, if the thing is true - and a
proof purely mathematical in content and formulation. This is what Godel
showed to be an impossible situation.

HRP: How has the role of logic in philosophy changed from the 1920s
and 1930s, when there was a lot of excitement about the foundational
role of logic?
Quine: 1 don't think the foundational role of logic has changed. But there's
been a lot of progress in specialized directions, with perhaps the greatest being
in axiomatic set theory and higher categories of infinity, and then also a
tremendous amount of progress in proof theory, of which Godel's proof is a
shining example.

HRP: You crowned your recent Pursuit of Truth (1987) with the motto,
"Save the surface and you save all." How have your empiricism and
behaviorism affected your view of what it means to give a philosophical
explanation?
Quine: That epigraph was accompanied with another one from Plato - "save
the phenomena." I find it particularly interesting that Plato appreciated this
attitude, but it's essentially just the statement of empiricism, the statement that
what we're trying to do is explain what we can observe. If we don't respect our
observations, and stick to them rather than revising them to fit our theory, then
we're not succeeding in our effort to achieve truth.
Behaviorism, as far as I'm concerned, is only an intersubjective empiricism.
It's empirical in attitude, but one doesn't settle, in the manner of Husserl and
the old epistemologists, for private, introspective data. When you take as your
data your own perceptions, and pool these with those of your colleagues, and
get the common denominator, then you have data which are pertinent to sci-
ence from the standpoint of intersubjective behaviorism. I don't see that as
going beyond what every modern scientist would subscribe to as a matter of
course.

HRP: In the
reduction of
common-sense
terms for intro-
spective human
states t o pub-
licly-accessible,
empiricist terms,
is there a danger
o f losing some-
thing? Can
empiricism d o
justice t o our
inner lives?
Quine: Here one
must distinguish two factors: the dreaming-up of hypotheses and the amassing
of evidence for them. All sorts of undisciplined thinking can be valuable as a
first step in thinking up bright ideas, way-out, highly imaginative, which turn
out to be just what's needed once you get down to finding logical connections
through experiments.
I don't think that this process neglects the natural mentalistic input - it
just means not settling for it. Daniel Dennett was very good on this topic in an
essay defending introspection, which took pretty much this line, as I read it.

HRP: You wrote in Pursuit of Truth that "there is nothing in linguistic

meaning beyond what is to be gleaned from overt behavior in observable
circumstances." Does the indeterminacy of translation hinder the inter-
subjective communication of scientific concepts, a communication which
seems to be important for the project of "naturalizing epistemology"?
Quine: No, I think not. The indeterminacy of translation doesn't get in the
way of translation - it allows that there are good translations and bad. It's
rather a question of the data that are available to see whether a translation is
good or bad. It proposes external, behavioral criteria for the test.
Here's the situation: you have two translation manuals that are both as +
good as can be. This is according to the empirical test of using them in normal
communication with a native, and finding that they lead to smooth dialogue
and successful negotiation, which could mean getting directions somewhere, or
the native's giving up. some priceless relic for some glass beads, or whatever
you're trying to bring about. Those are the tests.
If you have two manuals, arrived at independently, which both pass those
tests to perfection, then you try meshing them. Suppose it turns out that if you
use one of these manuals for a given text, it gives perfect, smooth translation,
and if you use the other one it works just as well, but when you try alternating
them on the same text it turns out to give incoherence. This illustrates what I
mean by the indeterminacy of translation. My conjecture is that [this situation]
is to be expected when we get into theoretical dialogue. However, it's not
something that would show up in entering into scientific communication; both
of these manuals are completely successhl on the empirical score. In that case,
the moral is: stick to the same manual.
But translation
isn't always possi-
1think 'tying up loose ends [of ble in science.
Today's physicist
science]' is a g o o d w a y of condensing talks about neu-
trinos, for
philosophy's purpose. " instance, and says
''neutrinos are
particles which
lack rest mass." That sentence certainly isn't translatable into the English of
1930, because there's no translation of "neutrino," not even a long paraphrase,
because "neutrino" has never been defined. None of the particles, not even
"electron," has ever been defined in the strict logical sense of definition. There
aren't even contextual definitions comparable to Russell's definition of singular
descriptions. No word or phrase, either by itself or in context, can take the
place of the word "neutrino" or the word "electron." This is no obstacle; peo-
ple use the word with no problem, and the native above can even use the word,
given a certain substructure of explanation. Explanation, but not translation.

HRP: Your view explains intersubjective communication on the basis of

readily observable events. But what happens when we're testing "manu-
als" about theoretical concepts or philosophical ideas? Presumably, such
manuals could conflict with each other, yet they don't necessarily allow of
observable evidence for the success of translation. This is more acute
given the potential existence of radically different conceptual schemes
among those communicating. What do we do in this case?
Quine: The way I see it, two different translators could in theory come up with
quite different conceptual schemes for the same people. I see this as a matter
of richness rather than poverty. It helps bring out the point that conceptual
schemes are almost completely a matter of human creativity - creativity to the
purpose, however, of matching up with the neural input. Theory is so over-
whelmingly much stronger and broader than the neural input that, of course,
you expect slack, and that's what makes me believe in the thesis of indetermina-
cy.
In the case we're imagining, we have two accounts of the natives' meta-
physics, which are not much alike. It is then a chastening experience to find
that they're equally faithful accounts. Now what there is to be communicated
of an objective, scientific kind is being communicated all the better, through
our appreciating the theoretical latitude.

HRP:Would it be fruitful for philosophers to become irritated about this

state of affairs and seek metaphysical explanations for this metaphysical
looseness?
Quine: I don't think so. The distinctive thing about science is that there are
checkpoints of observation. Everything that is compatible with those check-
points [is acceptable]. One is always trying, as Sir Karl Popper pointed out, to
"break" one's theory, devising the least hopeful experiment possible to subject
it to. That's what marks the difference between responsible science and fancy,
and it remains undisturbed through all this.

HRP:Your philosophy played a pivotal role in the demolition of the logi-

cal positivism of the 1930s. Do you feel any nostalgia for the philosophi-
cal optimism of those days?
Quine: I can see the attractiveness of that idea, but I also see something hope-
ful that seems to be taking its place: the tendency in certain philosophical cir-
cles (Dennett is again a shining example of this) to rub out, or at least blur, the
boundaries between philosophy and various sciences. Here, Simon Saunders
and others are trying to rub out the boundaries between physics and philoso-
phy, along with others like Abner Shimony. It's not only getting physics into
the philosophical circle, but philosophy into the physical circle - collaboration.
These people take serious, advanced seminars in physics. Roger Penrose, in The
Emperor's New Mind, is hoping that someone will come up with a new force, a
new particle, that will give us a more intuitive understanding of new finds in
quantum mechanics, of scientific concepts anyway. That's of as much interest
to philosophers as physicists, and the addition of philosophers trained in physics
might improve the situation in physics itself.

HRP:So the paradigm is not philosophy grounding physics, but rather a

mutual strengthening and smoothing between philosophy and physics.
Quine: Exactly. Furthermore, it's not just in physics. This effect is present in
the work of Dennett, with his seminars over at Tufts with people from neurolo-
gy, computer science, linguistics, and psychology participating. The bound-
aries between philosophy and all these fields are wavering and disappearing, and
this bears promise of a great new era.

HRP:Are we, then, on the verge of a new optimism about the role of phi-
losophy and science? Is there a feeling of possibility frnally "getting things
right"?
Quine: I don't know about getting things finally right - it's an interesting
question whether there's an end point, but I have n o firm opinion there.
There is certainly progress, that seems very clear. What excites me in this col-
laboration is the prospect of progress, not necessarily getting to the end of the
line.

HRP: What have you found to be the most exciting intellectual develop-
ment of the last 20 years?
Quine: Near philosophy, and of close philosophical relevance, I think of the
work of Hubel and Wiesel in neurology, and the utterly new picture that
they've given of the neurology of vision. It seems that it's not a matter of the
visual field being reflected isomorphically somewhere in the nervous system,
but a matter of various dominant features coming in separately from one anoth-
er, and not even being clearly synthesized until the moment of reflection. All
of this goes on in hundredths of seconds. It gives us a new angle of approach.
Reaching back before 1974, I think of advances in psychology of vision
made by Edwin H. Land, inventor of the Polaroid camera, and his theory of
color vision. There's no guessing what they're going to come up with next.

HRP: You seem to have considerable respect for the work being done in
cognitive science. How do you feel about the attitudes characteristic of
cognitive science - for instance, the conjecture that machines can dupli-
cate the intelligence of humans?
Quine: First, I find the Turing test for artificial intelligence unhopefd as a test,
because human behavior depends so much on everything that's gone o n
through many years, reaching back to infancy, and even to the distant history
of our corner of the world. Intelligence, it seems, is a matter of degree, and
I'm hindered by the fact that I have no satisfactory criterion for what consti-
tutes "thinking." When we just go by output, of course, computers are already
doing a quite remarkable job of thinking.
But I doubt that a machine could ever pass the Turing test. That was itself
an arbitrary criterion to separate thinking from computing, and I have no hope
for the development of a sharp criterion.

HRP: Famously, several of your papers written in the 1950s denied that
the analytic/synthetic distinction had any foundational value, at least in
the sense that Carnap and others wished. Does the distinction have any
methodological signXcance?
Quine: In The Roots of Reference I suggested a definition of analyticity which
seemed to me to approximate the layman's intuitive conception - not that the
layman uses the term "analytic." The analytic is what the layman calls "just a
matter of words." He dismisses someone's assertion as just a matter of words,
as a question of how t o use the words.
But we still want a criterion for analyticity, and one that has occurred to
me is that a sentence is analytlc (for a given native speaker) if he learns the truth
of the sentence in the course of learning one of its words. The obvious exam-
ple is "No bachelor is married." A native speaker clearly learns the truth of this
sentence by learning how to use the word "bachelor." A foreign speaker might
have learned the word rather by translation. This criterion covers not only all
sentences like that, but surely also all truths of elementary logic. Somebody
who affirms "p and q7' but denies "p" is, we say, misusing "and." H e hasn't
learned to use "and." That's true for all the basic principles of elementary logic,
and from the basic principles you get all of elementary logic.
Elementary logic, in the sense I intend it here, is complete. It covers
truth-hnctions, quantification, and identity - not set theory. This fits nicely
with what Frege, Carnap, and Kant all held. I don't think of the rest of mathe-
matics as being analytic. Furthermore, this isn't a concept that I can see apply-
ing across the board. This is what makes it unfruitful as a tool of philosophical
analysis. Nobody remembers how they learned the truth of each sentence, and
on the face of it,
it shouldn't mat-
ter. I think that there's no end t o the
S o a new
word, "momen-
important non-empirical beliefs and
turn," comes in. truths of history and sociolo~y,and per-
By definition, it's
mass times ve~oci- haps theoretical physics. You can add a
ty. questions whole bunch of them todether and they
there, an arbitrary
definition. We won't be enoudh to imply any observa-
dig up an
o b s o l e t e Latin
tions, and yet they're important ....
w o r d f o r this Sciencewouldbeparalyzedifwe
technical purpose.
But even so, rela-
excluded the untestables.
tivity theory
denies that momentum is simply mass times velocity. You have to get some-
thing in there about the square of the speed of light. So there's nothing about
analyticity that compels particular uses of words. That's why I felt that
Carnap's blanket application of the term in theoretical contexts generally was
obscuring the issues and getting on the wrong track. This matter of truth by
virtue of language does come in as a component, but not, I would argue, sen-
tence by sentence.
Clearly, verbal convention is a factor in the truth of any sentence. If the
syllables of the words in any sentence, however empirical it may be, were used
for another purpose, the sentence might be false instead of true. So the degree
to which language contributes to this or that sentence, this or that part of sci-
ence, can be methodologically and scientifically significant.

HRP: Is the gradation of analyticity and syntheticity in ordinary language

a philosophically interesting topic?What are some ways it might be talked
about?
Quine: I wouldn't think of it as coming into issue with particular words. But
sentence by sentence ...it seems reasonable that there would be gradations,
maybe consisting in the magnitude of the network of definitions that are neces-
sary to connect what we learn ostensibly with a sentence in question. That's a
highly theoretical question, and perhaps that would capture the gradation.

HRP: You gave us a sense in which elementary logic is analytic, but were
just now careful to exclude set theory from that sense, presumably because
of the implications of Russell's Paradox. What is the status of set theory,
and given its use in grounding the axioms of mathematics, where does
that leave arithmetic?
Quine: Russell's Paradox made all the difference. Frege thought the whole of
set theory was analytic, although even in the days before Russell's Paradox he
noted somewhere that the axiom that people would be most dubious about in
- -
terms of its ana-
lyticity was t h e
"[C]onceptual shemes are almost com- universal compre-
pletely a matter of human creativity - hension principle
[all predicates
creativitiy t o the purpose, however, of determine class-
es]. T h e ques-
matchin, up with the neural input. tion applies t o
Theory is so overwhelmingZy much higher m a t h e -
matics generally,
stron8er and broader than the neural from set theory
input that, of course, you expect slack, on up.
I see mathemat-
and that's what makes me believe in ics as sharing
the thesis of indeterminacy." empirical content
in a way connect-
ed with holism.
In general, if you're testing a scientific hypothesis, typically you're going to
need a whole cluster of other hypotheses before you get down to the level of
observables with implied consequences. Among the auxiliary hypotheses you'll
have to bring in, there will be some purely mathematical principles, differential
and integral calculus, and so on.
So you might say that in general a cluster of scientific truths and hypothe-
ses has an empirical "critical mass" if it is enough to imply observable conse-
quences. And often, to get critical mass you're going to need some purely
mathematical sentences. If you regard essential participation in a set of
hypotheses with critical masses as what it means to have empirical content, it
means that [all] applicable mathematics has empirical content. But it's not
empirical in the sense of John Stuart Mill; that is, arithmetic is not the product
of generalization from counting and counting again. Sometimes it is, I expect,
historically, but that's not the point.
Then I would apply this in particular to set theory. But this matter of hav-
ing empirical content must be separated from the old positivistic exclusion prin-
ciple of the meaninglessness of metaphysics. I think it's a mistake to require
that a sentence must have empirical content in order to be meaninghl. In fact,
I think that there's no end to the important [non-empirical] beliefs and truths
of history and sociology, and perhaps theoretical physics. You can add a whole
bunch of them together and they won't be enough to imply any observations,
and yet they're important. They seem plausible by virtue of symmetry, simplic-
ity, fitting-in to the things that we have well established by tests. These are
indispensable, moreover, in suggesting further hypotheses which one can test.
Science would be paralyzed if we excluded the untestables.

HRP:Within a "critical mass" of hypotheses in mathematics, exactly how

do the purely mathematical sentences relate to the empirical ones?
Quine: Here's something characteristic of mathematics, but perhaps not only of
mathematics: That these sentences are needed as components of sets with criti-
cal mass, but that
n o set purely of
mathematical
components has
critical mass. This
is a sense in which
mathematics can
be viewed as aux-
iliary t o science.
But then history
- something as
unlike mathemat-
ics as can be -
would also be
auxiliary in this
sense. Again, just
a set of historical
sentences about, say, ancient Rome, and nothing more, can't be tested without
the help of a lot more earthy generalizations. The tolerance that I'm urging
regarding sentences that don't, of themselves, have empirical consequences,
becomes curiously enough not merely a defense of highly speculative meta-
physics or of history, but of hard mathematics itself.

HRP:Your work is marked by a clear concern for style of expression.

What are your stylistic aims in writing philosophy?
Quine: Well, I certainly haven't thought about style enough to have principles
of consequence, but one thing that I tend to try to avoid is mutually conflicting
etymological metaphors.
One of the persistent errors that strikes me hnny, which I first saw in a
newspaper, and now it keeps popping up in print, is "stirring up tensions."
Another is "it was at the height of the Depression."
Another specific thing I dislike is the heterogeneity of languages in new
coinages. I like homogeneity of the components of a new word. For example,
a mathematician's "hypernatural numbers." No! "Supernatural numbers.''
Hyper- is Greek and natural is Latin. There's so much of that. I would hope
that a scientist who needs a new technical word for something would consult a
literate colleague, someone who knows Latin and Greek and could help him
get it right.
In more general terms, I strive for clarity and brevity. It happens so often
that by going over something and malung it shorter I improve it.

HRP:Where did you pick up your rhetorical principles and skills?

Quine: When I look back to my first book, A System of Logic (1934), and arti-
cles from those years, I find that it's a more stilted style - I wasn't as sure of
myself. My style's improved a lot since then. It's been mostly a matter of prac-
tice. I don't know of any source for my distaste for heterogeneities and so on.

HRP:At its best, it seems that American analytic philosophy tends toward
a radically clear,
taut style that
"One is always trying, as Sir Karl one sees particu-
Popper pointed out, to 'break' one's larly i n both
your work and
theory, devisin~the least hopeful experi- N 1 ,
ment possible t o subject it to. That's Goodman's. It
reflects a very
what marks the difference between different spirit
from the
responsible science and fancy. " German philo-
sophical prose of
someone like Heidegger. Do you t h i i that the tradition, and the values
of those who work in it dictate a particular style?
Quine: That's a very interesting point, and I believe perhaps it does. There are
certainly affinities between [my writing] and Goodman's. But when I think of
how clear and succinct Carnap's German was ...

HRP:Perhaps it's not national, but rather a divide between the analytic
and Continental traditions.
Quine: Yes, I think that's it. It's the same sort of taste that makes for mathe-
matical and logical elegance. It's linked to the desire for clarity and brevity.
But my interest in overtones in etymological metaphors would be an indepen-
dent factor. It comes from an interest in natural language as such that I've had
from way back; it just happened. I got into etymologies when I was still in
high school - I remember taking books out of the library about the origins of
words. I still have the same urge; 1'11 want to know what the origin of a word
is and have to look it up.

HRP:What modem philosophers do you follow with the most interest?

Who do you t h i i is on the right track?
Quine: I suppose that would have to include the work of my former student,
Donald Davidson. I don't agree with him on everything, but we're after nearly
the same thing, and along the same lines. We have very profitable discussions.

HRP: That's interesting - in a controversial reading of Davidson's phi-

losophy, Richard Rorty has taken Davidson as an exemplar of his own
type of ironist, relativist anti-representationalism. What do you think of
Rorty's stance?
Quine: I was surprised by reading Rorty's big book [Philosophy and the Mirror
of Nature], the one that made such a stir. He said very favorable things about
my philosophical views, which surprised me because I didn't like his position.
It strikes me as defeatist, and it shows a bit of the anti-scientific bias that is
more extreme in people like Derrida and Heidegger.

HRP:Those "Continental" thinkers are often distinguished from analytic

philosophers by saying that they do not take scientific inquiry as an intel-
lectual model. Your own project of "naturalizing epistemology," on the
other hand, reflects the conviction that scientific theory provides the para-
digm for what philosophical accounts should be like. Do you think it's
possible to do viable philosophical work which is organized around other
paradigms, like poetry or belles lettres?
Quine: It seems to me that the place where that sort of thinking has a function,
and is still within the domain of philosophy, is in philosophy of poetry and the
graphic arts. Whether we choose to call it philosophy or not, there is a place
for the strictly artistic use of language in conjuring up thoughts and visions
which help you get the point of an art work that you had missed before. I
don't think that this function could ever be managed by strict scientific prose.
But on the other hand, I don't see a place for it in scientific philosophy.

HRP:If a young philosophy student came to you for advice, and asked
where the most pressing work in philosophy currently lay, where would
you advise the student to turn?
Quine: One big project that has struck me as promising would be to take some
limited part of hard science, like Newtonian mechanics, and try to trace out
explicitly the logical connections of implication from its basic principles to the
observational checkpoints. In other words, an explicit amassing of evidence. It
would have the advantage not only of illuminating the epistemology of the sci-
ence, but also might even be a contribution to the science, by suggesting short-
cuts and simplifications, or by showing how some bit of theory just didn't get
used or serve its purpose. That method, if it succeeded with a sample chunk of
science, might spread to other cases, and, if it succeeded, might help make phi-
losophy a handmaiden to science, along with mathematics. cp