Kant versus Quine: Transcendentalism or

Naturalism?∗
Lars-Göran Johansson

1 Introduction
Kant wrote that Hume awaked him from his dogmatic slumber (1953, preface) Kant
refers to Hume’s arguments to the effect that we can give no justification of the idea
that there is a necessary connection between cause and effect and that this connection
could be known a priori. Kant accepts this negative part of Hume’s philosophy but he
rejects his positive doctrine, viz., that the connection between cause and effect is nothing
more than regularity and that the imposition of necessity is a projection into the world
from our minds. Hume’s explanation was that we are conditioned to expect the effect
to follow the cause, and this conditioned expectation make us think that ’now the effect
must come’ when we have observed an instance of the effect. This explanation does not
satisfy Kant.
Hume concluded from his association theory of the cause-effect relation that we
cannot justify the belief that external objects cause our impressions of them, because
the cause, i.e., the external object, is only available to us via its effect, the impression.
If cause-effect relations are associations between impressions, there cannot be any such
association, simply because there is only one impression. Hence we cannot give any
justification for our belief in an external world, i.e., our belief in independently existing
objects. In short, the entire project of giving an epistemological foundation for our
empirical knowledge was put in doubt by Hume. Hume’s way out, to reject the quest
for ultimate justification as an illegitimate request, was not accepted by Kant.
Kant’s response, in the first Critique, to Hume’s challenge was to distinguish be-
tween two levels of discourse, the empirical and the transcendental. The transcendental
discourse is an analysis of the conditions for the possibility of objective knowledge.
He found that these conditions are of two types. One is that our mind utilizes a set of
fundamental categories which are the most basic constituents of judgements, such as
∗
I would like to thank Antti Keskinen and Marcel Quarfood for several valuable comments on a draft
of this paper.

1
substance-accidence, cause-effect, contingency-necessity. The other type is our forms
of intuition (german ‘Anschauungsformen’), viz., time and space. Our minds are so
construed that we cognize objects as things in space and time, and in our judgements
we always utilize the fundamental categories. Hence, the forms of intuition and the
categories are in us, i.e., constitutive elements of our mind; hence these constitutive
elements determine the forms of all possible empirical knowledge, according to Kant.
From this two-level analysis follows a distinction between a phenomenon, a thing
as cognized by us, and a noumenon, a thing as it is in itself. This distinction should not
be thought of as a distinction between two objects, but rather two perspectives of one
and the same object. The famous phrase ‘Das Ding an sich selbst betrachtet’, the thing
as it is initself, is chosen so as to clearly indicate that it is ordinary physical objects we
are talking about; and the conclusion from the transcendental analysis is that we cannot
think of them without using our basic categories and forms of intuition.
If we would interpret Kant as saying that the phenomenon is distinct from the
noumenon and thus that the phenomenon is a representation of the noumenon, we would
attribute to Kant a representational theory of perception, and that was precisely what
Hume had shown could not be justified; Kant’s transcendentalism is thus not a theory
about how we represent external objects but an attempt to do without representations.
So we cannot really say anything at all about things as they are in themselves, ac-
cording to Kant. In fact, we cannot even say about them that they cause our perceptions;
for the concept of cause is one of the categories and using it we are in the empirical dis-
course. It is no wonder that some of Kant’s followers dismissed the notion of things
as they are in themselves as having no theoretical role to play; thus they became full-
blooded idealists. All there is are phenomena, objects as perceived. But then, we are
back to Berkeley’s position; physical objects are nothing but collections of perceptions.
But this conclusion flies in face as an absurdity. So what are the remaining options?
One is to become a metaphysical realist; another is to stick to Kantian transcendental-
ism; but a third option has attracted more and more philosophers the last thirty years or
so; naturalism.

2 Naturalism a la Hume and Quine

One response to Hume’s challenge, or rather a development of one strand of his thoughts,
is to become a naturalist. In my view, the fundamental idea in epistemological natu-
ralism is to dismiss the request for a philosophical, ultimate justification of empirical
knowledge, but instead take our most basic beliefs as a starting point; not as an indu-
bitable starting point, but as a beginning. Hume indicates his version of naturalism as
follows:
Thus the sceptic still continues to reason and believe, even tho’ he asserts,
that he cannot defend his reason by reason; and by the same rule he must

2
 assent to the principle concerning the existence of body, tho’ he cannot
pretend by any arguments of philosophy to maintain its veracity. .... We
may well ask ’What causes induce us to believe in the existence of body?’
but ’tis in vain to ask whether there be body or not? That is a point, which
we must take for granted in all our reasonings. (1985, 238)

One could thus describe naturalism as a rejection of the Cartesian goal of establish-
ing a first philosophy, an a priori justification of all knowledge. This is also what Quine
means by naturalism ( Quine 1976).
Naturalism, in the sense of Hume and Quine, thus dismisses the idea of an ultimate
foundation for all knowledge, but not that specific knowledge claims must be justified.
In Quine’s version, as in many others, this leads to a form of restricted coherentism: a
specific statement is justified by other statements, and these in turn by other, but there is
no self evident basis. The coherentist trend in Quine’s philosophy is restricted by taking
observation sentences as starting point. Quine would not call them ‘self-evident’; his
view is rather that we assent, or dissent, to them directly, without much ado, and we
are hardly ever prone to argue about them. Quine’s view is analogous to Hume’s; there
are some things we simply accept without raising the question of justification. But
other statements can be revised in the light of further observations and demands of
coherence, simplicity etc. In Quine’s earlier philosophy (Quine 1951) even logical and
mathematical statements were revisable, whereas in his later writings (xxx) he has called
this position ’legalism’, thus indicating that strictly speaking it is true, but in practice
we would hardly in any circumstances give up central beliefs in mathematics and logic
when confronting problems of bringing together theory and empirical evidence.
Quine describes the relation between epistemology and empirical science as ‘mu-
tual containment’: epistemology is part of science, ’a chapter of cognitive psychology’;
but it is also the case that epistemology contains many parts of science, such as neuro-
science, psychology and set theory. (Quine, xxx)
Another strand in Quine’s philosophy is his hostility towards intensional linguistic
contexts, or more precisely, his dismissal of objects of propositional attitudes; thoughts,
desires, willings, beliefs, propositions etc.; they are not Quine’s cup of tea. Many
philosophers, while agreeing in rejecting the goal of a first philosophy, think this is
wrong and against a true naturalistic view-point; It belongs to our nature to describe
ourselves and others as having beliefs, thoughts, desires etc,, to say that these states
are directed towards objects and that they have contents. These states of affairs are just
as natural as physical states of affairs. (This was certainly Hume’s view.) However,
Quine’s rejection of intensional objects is not part of his naturalism. His reason for dis-
missing intentional contexts is that quantifying into them, and thus to accept intentional
objects in the ontology, conflicts with the principle of extensionality and this principle
is not motivated by naturalism. So dismissing mentalistic language, and intentional ob-
jects in general, is, for Quine a doctrine that is independent of naturalism. However, this

3
is beside the topic of the present paper. In passing it should perhaps be noted that Quine
does not think that ordinary talk about thoughts, wishes, sensations etc, is illegitimate;
only that these objects are not suitable for philosophical and scientific discourse.

3 Naturalism and the transcendental mind

How, then, should a naturalist respond to the Kantian ideas (i) that there are conditions
for the possibility of knowledge, (ii) that these conditions are in us, that they can be
deduced by a study of the constitution of our mind and (iii) that the mind is outside the
empirical world, i.e.that the mind is transcendental.
The first two of these Kantian theses are not contrary to naturalism or empiricism.
The conditions for the possibility of knowledge is, in the naturalist view, determined
by a natural information process that begins in the external world and ends with an act
of perception in our mind. And empirical study of our sense organs reveal that they
are not passive registrators of incoming signals; the sense organs and our mind actively
process incoming stimuli; some are selected, some are suppressed and the selected ones
are organised into certain patterns. Hence, the activity of our mind and its constitution
might very well contribute both to the structure and to the content of our perceptual
knowledge. But what is contrary to naturalism is the Kantian conception of the mind as
something transcendental, something outside the empirical world. The mind, no matter
in what terms we analyze its operations and constitution, is, according to naturalism,
part of the empirical world. So the basic invention of Kant, to distinguish between two
levels of discourse, the empirical and the transcendental, conflicts with naturalism. It
is precisely for this reason I prefer naturalism, since I have doubts about any a priori
inquiry into our minds.
Transcendental inquiry is an a priori inquiry, a kind of introspective investigation
of the operations and constitution of the mind. But what is mind? I think we have
strong empirical reasons to hold that its constitution and operations are heavily depen-
dent on what happens to our body; the mind is not something floating free above the
empirical world. In fact I think the only reasonable stance is some kind of monism, i.e.,
that mind and body are not two different objects or substances, but fundamentally one
thing, described by different kinds of concepts. Roughly, we use physical, biological or
functional concepts when talking about ourselves as a body, whereas we use intentional
language when talking about ourselves as minds. Accepting this does not entail reduc-
tionism, as is well known, witness Davidson’s anomalous monism (Davidson, 1980).
Quine fully embraces this view, see his (1990, §29) and (1995, 87-88). So the first argu-
ment against transcendentalism is that its object, i.e. the mind, is in fact not something
outside the empirical world. It is something in the world and the naturalist view is that
its exploration should be done by ordinary empirical methods.
Secondly, one may reasonably doubt the reliability of introspection; we have since

4
long understood that even the most honest person may completely fail to observe her
mental state. This is a lesson we have learnt from the failure of 19th century empirical
psychology (Watson, 1913), Freud, and modern cognitive science. Kantians may say,
ok, that’s correct but beside the point, for transcendental inquiry only concerns struc-
tural features of our mind. Then the sceptic naturalist will ask: even if we grant that
the transcendentalist is correct in describing the structure of his own mind, what is the
reason for saying that all minds have the same structure? Surely the transcendentalist
can look only into his own.
These doubts are not only directed towards transcendentalism but to any philoso-
phy which takes for granted that we all have direct and certain knowledge about our
thoughts, perceptions and judgements. An empiricist might reasonably say: these inner
things are not publicly observable. We cannot say that we have knowledge about them
since private introspection is uncertain.
The linguistic turn in philosophy is a response to such concerns. For judgements,
volitions, sensations etc, are only made subject for discourse by being expressed aloud,
and then we have something intersubjective to begin with: statements made by people.
And now we can perform something analogous to transcendental inquiry into the condi-
tions for knowledge, viz., a logical analysis of the expressions for judgements etc., viz.,
declarative sentences. The naturalist may in a similar fashion as the transcendentalist
say: there are conditions for knowledge, viz., the ability to use language for expressing
declarative sentences and an inquiry into the general features of declarative sentences
reveal those conditions.
Quine begins his discussion of these matters (1960) with language learning and
describes the interaction between child and caretaker in a common environment without
using any mental concepts. The child is rewarded when uttering single words in the
correct circumstances. These single words may not, at the beginning, be interpreted as
that the child is naming things, but rather as use of one-word sentences. For example,
when the child says ’mama’ in the presence of her mother, this should not be taken as
proof for the the child has discerned a certain person; we could just as well interpret
the word as ’it is mummying’ or ’I want it to mummy’. It is only after the child has
developed rather complex linguistic abilities, such as use of subjunctive clauses and so
called essential pronouns that we have evidence enough for saying that the child refers
to objects (1995, 26-27)
Maddy (2007, ch. III.5) critizes Quine on this point. Here argument is that recent
research in cognitive psychology (e.g. Spelke et.al, 1995, Xu, 1997) has obtained evi-
dence that small children, long before they begin to talk, discern and recognize visual
objects. So when the child later begins to utter one-word sentences, we may assume
that the function of these words is that of naming, as signalling that the child discerns
an object, even before it has developed the full referential apparatus of language.
Let’s grant Maddy’s point, for the sake of argument. Still, I don’t think this under-

5
mines Quine’s point, since I take it to be fundamentally epistemological; just because
we hear the child using a word which adults use referentially, we are not justified in
concluding that it uses it the same way. Quine’s point with the discussion about child’s
language use is the same as with his famous gavagai example; when hearing some-
one saying something we cannot infer the speakers ontology without using a translation
to our own language. Maddy further argues that these results from research in child
psychology give us reason to hold that the natural world in fact is composed of inde-
pendently existing objects, such as animals, trees, houses, humans, etc. Here argument
seems to be based on an evolutionary perspective; our human cognitive apparatus has
during evolution developed by adapting to the structure of an independently existing en-
vironment. We perceive, basically, physical objects, and hence we have reason to think
that these objects actually exist out there in the physical world.
Now, an adherent to Quine’s philosophy may argue that already a six month’s baby
has developed a lot of cognitive abilities during its interaction with her caretakers; hence
there are reasons to hold that the child’s structuring of the environment into individual
physical objects having a certain permanence in time is driven by practical needs, not
by pre-existing ontological structures. Our cognitive apparatus, just like that of other
animal’s, is basically constructed so as to be goal-driven. I’ll return to these matters in
the next section.
One of Quine’s conclusions is that a declarative sentence in ordinary talk is com-
posed of a general term and a singular term. We are prone to say that this structure
reflects the structure of our judgments, thus assuming that first we form a judgement by
applying a concept to an object and then we construe a sentence expressing this judge-
ment. Prima facie, it seems reasonable to assume that judgements come first and that
sentences, if they correctly express our judgement, are construed to reflect this structure.
But I think there are no good reason to assume that people can make judgements before
and independently of mastering a language; in fact there are good empirical reasons to
hold that the ability to make judgements and the ability to use linguistic signs for com-
munication go hand in hand. (Gärdenfors, 2000 and references therein). And Quine,
who does not accept the mental as an independent realm, would agree of course.
The basic epistemological point is that our utterances, whatever we talk about, are
clearly and easily available, whereas the activities of the mind, at best, are inferred from
publicly available evidence. This is the fundamental reason behind the linguistic turn in
philosophy 100 years ago.

3.1 The relation between the cognition process and its object
Kant’s position is often described as the combination of transcendental idealism and
empirical realism. What does that mean? I think the best way to understand the distinc-
tion between realism and idealism is in terms of the relation between the act of cognition
and the cognized object. This is how Fichte viewed the matter and it is Göran Sundholm

6
who brought my attention to Fichte’s view-point:

”In his characterisation of the realism/idealism (antirealism) debate Fichte

noted that basically there are only two epistemological options. The posi-
tions may by formulated in terms of the act/object dichotomy:

act
− − − − − − − − − →object

Either you determine the object of knowledge as the object of the act, and
then you are an idealist, or you determine the act in terms of a prior object
towards which the act is directed, and then you are a realist (or dogmatist,
as Fichte said, being an idealist himself). If Fichte is right in this (and I sus-
pect he is), the point is moot whether there is a neutral background position
from which the issue between the realists and idealists can be adjudicated.
For Fichte, it is clear that there is not. If this be so, it would serve to ex-
plain why the realism/idealism debate so often makes a futile impression;
in place of a clear-cut decision, we find endless refinements of positions
into sterile scholasticism, and conversions from one side to the other rarely
takes place.” (Sundholm, 2006)

This characterisation seems to me correct and informative. So Kant’s transcendental

idealism consists of two theses; i) the object of cognition is constituted by the cognition
act; that’s why we cannot say anything at all about the thing in itself, i.e, independently
of our cognitive acts; ii) we are able to formulate this position only at a level of discourse
where we reflect on the necessary conditions for objective knowledge, i.e., when we
perform a transcendental inquiry. Empirical realism is the thesis that once the act-object
is constituted, it is an empirical matter what properties it has. In the empirical process
properties of the object is discovered; it is the object that determines the results of the
cognitive act.
This is Kant’s response to Hume’s scepticism regarding the possibility of knowl-
edge about external objects; it is possible to know about empirical objects, because the
very structure of our cognitive processes, as it were, shapes them. The conclusion is
that we cannot even say about the external world that it is constituted of objects with
properties; it is we, in our cognitive activities, who structure the sense impressions in
that way. Thus, to repeat, Kant rejects the traditional notion that we have in our mind
representations, ideas, of external objects; the distinction between objects and ideas of
objects is a mistake.
Now, I find it rather peculiar that one may arrive at a position quite similar to this
form of idealism from a naturalistic point of departure. That is exactly what Quine did.
Quine’s account of cognition and perception is entirely naturalistic; he begins by
considering our empirical knowledge about sense perception, as viewed from outside;

7
that is, he does not avail himself of any mentalistic notions. So he considers how stimuli
arrive at our sense organs (he often talks about ’triggering our receptors’) and how these
somehow selects salient (for us!) features in the surroundings, i.e., features that we
become aware of. In this process we construct objects in the sense that we discern a
part of the visual scene an focus attention on it as a unit, whereas other parts become
background. This is an active process of our nervous system/mind. (Kant’s counterpart
to this constructive process resulted in ’the transcendental unity of apperception’)
An object that we observe is a unit of perception. This step from a stream of stim-
uli to an object could be done in different ways, a conclusion that Quine draws from
his famous discussion of the gavagai-example. The field anthropologist visiting a tribe
with unknown language hears someone saying ’gavagai’ when a rabbit passes the scene.
How may he interpret this sentence? Quine discerns three different but equally admis-
sible interpretations: ’Lo, a rabbit.’ Lo, a timeslice of rabbithood’, or ’Lo, a set of
undetached rabbitparts.’ The point is that by observing speech behaviour and being in
the same environment as the speaker is not sufficient for determining the speakers ontol-
ogy. Nothing of what we can observe, including the actions of the speaker, is sufficient
to determine which of these three ways of structuring the incoming stimuli is actually
going on in the speaker’s mind.
Quine’s point is epistemological; just because we see a rabbit and hear a sentence,
we cannot decide the ontological structure of another speaker’s thoughts or language.
But we can observe dissent or assent to something said; hence what can be observed is
the attitude towards the uttered sentence and it is easy to see that the same applies to
speakers of our own language. From this Quine infers that the semantic counterparts to
assent and dissent, truth and falsity, are the primary notions in semantics. In contrast to
the majority of semanticists, Quine does not explain truth in terms of reference, satis-
faction and meaning; he reverses the direction and explains reference in terms of truth
(and he endorses, of course, the deflationist view of truth). ’Meaning’ he dismisses as a
mentalistic notion.
A speaker can be observed to assent to a sentence. We interpret the sentence in
our language, thus providing it with a structure, in the simplest case as constituted of a
singular term and a general term. If the speaker assents to the sentence, it follows that he
holds that the singular term of that sentence refers to an object. But which objects does
he really assume? Because of the possibility to translate his sentence differently, with
different ontologies, we cannot decide that independently of the chosen translation. But
this is not a mere shortcoming of our ability to translate. Quine claims, again and again,
that there is no fact of the matter which translation is the correct one. Translations are
not uncertain, but indeterminate. It follows that ontology is theory-relative.
The ontology comprises those objects that in that theory are accepted as values of
variables; the things that are assumed when using expressions like ’some time slices of
rabbithood....’ or ’All rabbits....’ As Quine puts it, to be is to be the value of a variable.

8
 Quine arrives at a similar conclusion about theoretical objects, i.e., objects postu-
lated in our scientific theories. Here he uses logical arguments. Suppose a scientist
states the sentence Fa, i.e., he claims there exists an object a satisfying the predicate F.
Another scientist may translate his statement using so called proxy functions, i.e. map-
pings from one category of objects to another category, and at the same time reinterpret
the predicate; so a is F becomes proxy-a is proxy-F. Both theories have exactly the same
empirical consequences, viz., they entail exactly the same set of observation categori-
cals (Quine, 1981,19). One may think about this as merely two different formulations
of one and the same theory, in particular if the proxy functions can be explicitly given.
So objects varies, but theory is the same.
Some would instead say that we have two different theories which are underdeter-
mined by evidence, in fact by all possible evidence. For Quine this doesn’t matter, it is
only a matter how we chose to individuate the general term ’theory’ and this is of little
importance.
His conclusion is that the choice of objects has no epistemological significance,
neither in theoretical science, nor in ordinary talk, contrary to what most philosophers
think:
”Our talk of external things, our very notion of things, is just a conceptual
apparatus that helps us to foresee and control the triggering of our sensory
receptors in the light of previous triggering of our sensory receptors.” (1981,
p.1)
And later in the paper:

”Structure is what matters to theory and not its choice of objects.........I ex-
tend the doctrine to objects generally, for I see all objects as theoretical”
(1981, p. 20)

And still later:

”We must speak from within a theory, albeit any of various. Transcendental
argument, or what purports to be first philosophy, tends generally to take on
rather this status of immanent epistemology, insofar as I succeed in making
sense of it. What evaporates is the transcendental question of the reality of
the external world - the question whether or in how far our science measures
up to the Ding an sich. (1981, p. 22)
So I think it is quite clear form these quotations (there are many more indicating this
view) that the objects we observe and talk about in science and in everyday life are,
according to Quine, constructs, vehicles that help us in forming efficient expectations.
The basis is assent and dissent to observation sentences. If we transform Fichte’s de-
scription of the idealism/realism dispute to a linguistic framework, we can say that, for

9
Quine, the act of assent or dissent, which is observable, is primary to the object talked
about.
Quine and Kant thus agree that questions about how the world is in itself indepen-
dent of our ways of cognizing and using language cannot be asked, for such questions
presuppose an impossible stance, i.e., that we can have knowledge without applying the
structural features of language use and thinking. But Quine, unlike Kant, dismisses the
quest for a rational basis for empirical science; none is needed according to Quine.
The quest for safe foundations is triggered by sceptical arguments and Kant wanted
to rebut Hume’s and others scepticism by showing it has ramifications. But Quine thinks
transcendentalism, or any other foundational theory, is an overreaction to sceptical ar-
guments. Once we realize that scepticism regarding any concrete statement is motivated
by something we think is safer than the doubted statement, and that this could in turn
be doubted, we are in an infinite regress.1 But why this urge for complete certainty? In
our everyday life, as in science, we are satisfied with less than certain knowledge and
we are, at least in our philosophical mood, prepared to reevaluate most of our beliefs.
So why not begin by the most plausible things and be prepared to reevaluate any partic-
ular statement if new evidence comes up? This is naturalism+ fallibilism. In fact, the
urge for complete certainly is devastating; the chain of justification demands cannot be
stopped.
So idealism, in the sense that objects of cognition are secondary to the cognition act,
is (also!) a consequence of a naturalistic approach in epistemology, provided we accept
that cognition and language use are two sides of the same coin. So both Kant and Quine,
astonishingly, arrive at the same stance in ontology, albeit via quite different routes.
But, the reader might say, hasn’t Quine often expressed what he calls ’robust and
stubborn realism’; he claims to believe in sticks and stones, physical objects of various
kinds. Doesn’t this conflict with viewing them as theoretical constructs? No. Saying
that objects are theoretical constructs doesn’t entail that they do not exist, neither that
they exist only in our minds, quite the opposite: for when using general terms with
divided reference (’chair’, ’electron’, ’government’ etc.,) we thereby impose criteria for
individuation and identity among the things the general term is true of and then we can
quantify in the domain; we are entitled to use expressions like ’All governments’ or
’some chairs’. Having done that we have committed ourselves to the existence of those
objects we quantify over, expressed by Quine with the famous phrase, ‘to be is to be the
value of a variable’. And, of course, Quine’s idealism is totally different from classical
idealism a la Berkeley: The classical idealism presupposed a mind-body dichotomy,
which is incompatible with Quine’s naturalism and extensionalism.
1
Some sceptics have, from the plausible statement that any statement can be doubted, drawn the
conclusion that nothing can be known. This is a logical fallacy, viz, scope inversion of the universal
quantifier and the modal operator; From ’Any p can be false’ it does not follow ’It is possible that for
every p, p is false’. see Davidson (2001, p.194)

10
 The difference between metaphysical realism and Quine’s realism is that the meta-
physical realist thinks that the natural world is composed of individual objects, with
different properties, independently of us humans, whereas Quine, rejects this notion as
incoherent; the splitting up of the natural world into individual objects and classifying
them according to certain criteria is something we do for certain purposes. But once
this is done, i.e., once we have learnt language, we have accepted those objects that are
assumed as values of variables as real.
But, the realist may say, I grant that the only evidence we could have for the struc-
ture of the external world is empirical evidence, which does not include introspection,
and I grant that our cognitive acts structure the world into relatively permanent physical
objects, bodies. This is a fact about us humans. But I have an explanation of these facts,
viz., that biological evolution has made us such. In other words, of the possible ways our
cognitive organs may process incoming information, natural selection has chosen one as
optimal, viz., that of the world consisting of relatively permanent objects moving around
and changing properties and this choice is optimal in the biological sense of being ben-
eficial for survival and reproduction. But why is it optimal? The best explanation is
that that’s how the world actually is. Moreover, empirical evidence from cognitive psy-
chology, in particular studies of children before they master language, reveal that they
at least from 4-6 months age, cognize bodies in their immediate environment. (Maddy,
2007, pp. 245-58). So there are good reasons to assume that there exists in the world
relatively permanent physical objects independently of us.
I strongly suspect that Quine would resist any such conclusion, for his basic point
remains: we can ride a proxy function from a preferred ontology to some other, never
heard of set of objects, while all empirical evidence, are the same. That we humans all
are alike in our cognitive interactions with our environment in cognizing visual bodies
doesn’t prove that this is the metaphysical structure of the world. Quine could use
van Fraassen’s (1980, 20) reply in the realism-antirealism debate. Scientific realists
typically hold that the best explanation of the success of modern science is that our
theories are approximately true. van Fraassen replies that the best explanation is that our
theories are empirically adequate; nothing more could be inferred because all theories
are underdetermined by all possible evidence.

4 Scientific realism and metaphysical realism

Quine and Kant insist that we cannot step outside ourselves, as it were, and look at
the world from a view from nowhere, a viewpoint in which no conceptual apparatus
is presupposed. To think is in Kant’s view to use concepts. Quine, reluctant to talk
about mental activities, instead talks about using langauge, holds that talking is to use
predicates with criteria for satisfaction, but one may assume that thinking and talking
is structurally similar activities. Asking what there is from a perspective from nowhere,

11
a perspective which does not presuppose things that are identified as particulars and
subject to predication (Quine) or falling under concepts (Kant) is impossible.
It is strongly tempting to say that that’s how the world must be like, divided into
particular objects with different properties and that modern science has, in the main,
revealed that there are objects, (and we know which!) and properties independently of
our theories and ways of thinking. But when we express this thought we use exactly this
structure of singular terms referring to objects which are subject to predication; when
making this assertion we presuppose what is argued for.
We cannot, for sure, say that all theories are equally good or that all ontologies
are admissible. But the only way a false theory may display its falsity is in entailing
predictions that come out false. If so happens and we dismiss the theory, we were
wrong about what exists, provided the better theory has a different ontology. The only
way we can be proved wrong about what exists is by a mismatch between our theory
and its observable consequences. There is no other way saying what there is than saying
’there are objects such and so’.
From this account of Quine’s philosophy it should be clear that Quine endorses Rus-
sell’s theory of definite descriptions. But this theory is not universally accepted. Adher-
ents to non-descriptivism holds that there are at least some objects that are not identified
descriptively; some objects can be directly referred to without using any description.
Direct reference theories have two aspects: one is the original baptizing of the object,
the other is, to use Putnam’s term ’the linguistic division of labour’. Other persons using
the term refer to the same object as the baptizer and may do so without having any true
beliefs about it. This seems plausible, but it is irrelevant for the present discussion. It
is the other aspect that is crucial; What is involved in the original baptism, one person
deciding to name an object (or a substance) by a chosen singular term?
Well, the crucial point is how this question is phrased. We observe that in non-
descriptivist theories it is taken for granted that the object is there, ready to be observed
in front of the baptizer before any cognitive acts have occurred. Suppose this is true and
suppose the baptizer discerns it and gives it a name. But what is involved in discerning
an object from the rest of the scene?
Discerning an object and giving it a name means that we have identified it as an
object for predication. It can be talked about and non-descriptivists go as far as holding
that it is possible that a speaker may have only false beliefs about it and still refer to it.
That means that the object talked about must satisfy an identity criterion, albeit speakers
using the name in general may not be aware of this criterion. It suffice, according to
non-descriptivists, that speakers share the intention to talk about the same object. In
short, non-descriptivists take for granted that objects exists out there independently of
our ways of cognizing them and talking about them.
It is thus clear that non-descriptivist theorists take for granted precisely what Quine

12
and Kant denies, that is to say, they presuppose metaphysical realism.2 Quine could
simply say about this original baptizing: The same point as with ’gavagai’ applies even
if we speak the same language; indeterminacy begins at home and there is no escape
from descriptivism, since the object you identify is constituted by a description, i,e,
your choice of general term and it annexed principle of individuation. Even if you use a
name for the object, you presuppose there is a fact of the matter whether it is the same
thing or not you observe at another occasion and this fact must be expressed in terms of
a description of the object.3
Kant held that the conditions for the possibility of knowledge are in our minds;
Quine may properly be described as holding that the conditions for the possibility of
knowledge lie in the use of language. ( Both Kant and Quine would certainly accept that
they have restricted the term ’knowledge’ to mean discursive knowledge, not practical
abilities.) But the ability to use language is an empirical fact about human organisms;
so for Quine the conditions for the possibility of knowledge are not transcendental; they
are natural.
Russell became a realist as a reaction against Hegelianism, prevalent among his
teachers during his first years at Cambridge (and so did Moore). It is slightly ironical
that Quine, who began his career by writing a dissertation on Principia Mathematica and
who shares many of Russell’s philosophical convictions, should end up as an idealist.
Russell wrote (1956, p. 2) that Queen Victoria was ’not altogether sympathetic’ when
hearing from Russell’s grandfather that a monarch should be dismissed if he proved un-
satisfactory. I guess that Russell similarly would have been ’not altogether sympathetic’
if he had lived long enough to hear that his great follower becoming an idealist as a con-
sequence of his application of the theory of definite descriptions and a thoroughgoing
empiricism and naturalism.

References
Davidson, D. (1980). ’Mental Events’ pp. 207-228 in his Essays on Actions and Events.
Oxford: Oxford University Press.
Davidson, D. (2001). ”Epistemology Externalized”, pp. 193-204 in his Subjective, In-
tersubjective, Objective. Oxford: Oxford University Press.
Hume, D. (1969). Treatise of Human Nature. London: Penguin.
Kant, I. (1953). Prolegomena to any future metaphysics that will be able to present itself
2
This point is intimately connected to the debate about the viability of modal logic; Quine’s hostility
against modal logic is based precisely on his view in ontology. This point is not very well known, but
Antti Keskinen has clearly seen this, see his Quine’s Critique of Modal Logic and his Conception of
Objects
3
There is one alternative for non-descriptivists, viz., to refer to ’primitive thisness’, haeccity, as con-
ferring identity to objects, but I think this is too desperate for most of them.

13
as a science. P Gray-Lucas (ed). Manchester: Manchester University Press.
Kant, I. (1965). Immanuel Kant’s Critique of pure reason. (Unabridged ed.) New York:
St Martin’s Press.
Keskinen, A. (2010). Quines Critique of Modal Logic and his Conception of Objects.
University of Tampere.
Maddy, P. (2007). Second Philosophy. A naturalistic method. Oxford: Oxford Univer-
sity Press.
Niinilouto, I., Sintonen, M. and Wolenski, J. (eds.) (2006). Handbook of Epistemology.
Dordrecht: Kluwer
Quine, W.V.O (1969). ’Epistemology Naturalized’ pp. 69-90 in his Ontological Rela-
tivity and other Essays. New York: Columbia University Press.
Quine, W.V.O. (1981). ’Things and Their Place in Theories’, pp. 1-25 in his Theories
and Things. Cambridge, Mass: the Belknap Press of Harvard University Press.
Quine, W.V.O. (1990). Pursuit of Truth. Cambridge, Mass: Harvard University Press.
Quine, W.V.O. (1995). From Stimulus to Science. Cambridge, Mass: Harvard Univer-
sity Press.
Russell, B. (1956). Portraits from Memory. New York: Simon and Schuster.
Sundholm, G. (2006). ’Antirealism and the Roles of Truth.’ In Niinilouto, Sintonen and
Wolenski (eds).
Spelke, E., Gutheil, G. and Van de Walle, G. (1995). ’The development of object per-
ception’ in S. M. Kosslyn and D. N. Osherson eds., Visual Cognition: An invitation to
Cognitive Science, vol 2., Cambridge, Mass: MIT, pp. 297-330.
Watson, J.B. (1913). ”Psychology as the Behaviorist Views it.” Psychological Review
20, 158-177.
Xu, F. (1997). ’From Lot’s wife to a pillar of salt: evidence that physical object is a
sortal concept’, Mind and Language 12, pp 365-392.

14