AGAINST “OSTRICH”

NOMINALISM:
A REPLY TO MICHAEL DEVITT
BY

D. M. ARMSTRONG

1)
X AM dissatisfied with my treatment, in Volume I of
Universals and Scientific Realism (Armstrong, 1978), of what 1 there called
“Ostrich” Nominalism. Michael Devitt’s vigorous defence of Quine, whom I
accused of being such a Nominalist, gives me a second opportunity. (I should
like to thank Devitt for comments on earlier drafts, and for the pleasant spirit
O um em muitos não prova,
in which this controversy has been conducted.) segundo A. ,a existencia de
universais, mas é um um forte
argumento preliminar para
I. Quine and the “ One over M any” acetitação destes

2)
I think that the main argument for the existence of universals is Plato’s “One
over Many.” I do not think that it proves straight off that there are universals.
But I think that it shows that there is a strong preliminary case for accepting
universals. There are various sorts of Nominalists (I spoke of Predicate, Concept,
Class, Mereological and Resemblance Nominalists) who seem to perceive the
strength of the “One over Many” but who maintain their Nominalism never­
theless. There are, however. Nominalists who deny that the argument has any
force. These I christened, tendentiously enough. Ostrich Nominalists. Quine is
certainly one who denies the force of the “One over Many.”
3) In Chapter I of From a Logical Point of View (1953), that is, the well-known

paper “On What There Is,” Quine makes a philosopher whom he calls ‘McX’’
advance the “One over Many” :

4) Speaking of attributes, he [McX] says: “There are red houses, red roses and red sunsets;
this much is prephilosophical common sense in which we must all agree. These houses,
roses and sunsets, then, have something in common; and this which they have in common
is all 1 mean by the attribute of redness.” For McX, thus, there being attributes is even
more obvious and trivial than the obvious and trivial fact of there being red houses, roses,
and sunsets, (pp. 9 -1 0 .)

5)
In my view, Quine has here made McX considerably overplay his hand. I
would wish to start in a much more cautious way by saying, as I say on p. xiii.

Pacific Philosophical Quarterly 61 (1980) 440-449. 0031-5621/80/0400-0440$01.00

Copyright © 1980 by University of Southern California

440
 AGAINST “ OSTRICH” NOMINALISM 441
6)

that:

. . . many different particulars can all have what appears to be the same nature.

and draw the conclusion that, as a result, there is a prima facie case for postulating
universals.
7) Quine, I think, admits or half-admits the truth of this premiss, though in a
back-handed way, when he says, in the course of his assault on McX:

One may admit that there are red houses, roses and sunsets, but deny, except as a popular
and misleading manner of speaking, that they have anything in common, (p. 10.)

8) Quine here allows that there is a popular manner of speaking in which different

red things are said to have something in common. But he does not seem to
realize just how ubiquitous such manners of speaking are. We (that is, everybody)
are continually talking about the sameness of things. And most of the time when
we talk about the sameness of things we are talking about the sameness of
different things. We are continually talking about different things having the
same property or quality, being of the same sort or kind, having the same nature,
and so on.
9) Philosophers have formalized the matter a little. They draw the enormously

useful Peircean distinction between sameness of token and sameness of type.

But they are only formalizing, making explicit, a distinction which ordinary
language (and so, ordinary thought) perfectly recognizes.
10) G. E. Moore thought, correctly I believe, that there are many facts which even

philosophers should not deny, whatever philisophical account or analysis they

gave of these facts. He gave as an example the existence of his hands. We can
argue about the philosophical account which ought to be given of material
objects, such as Moore’s hands. But we should not deny that there are such
things. (He was not arguing that their existence was a logically necessary or
logically indubitable truth.) I suggest that the fact of sameness of type is a
Moorean fact.
11) Any comprehensive philosophy must try to give some account of Moorean

facts. They constitute the compulsory questions in the philosophical examination

paper. If sameness of type is a Moorean fact, then, because Quine sees no need
to give an account of it, he is refusing to answer a compulsory question.
12)
Here is one answer to the question. When we speak of sameness of token,
the sameness of the Morning and the Evening star to coin an example, we are
speaking of identity. But when we speak of sameness of type, of two dresses
being the same shade of colour for instance, sameness is merely a matter of
resemblance (on one view between the dresses, on another between two property-
instances). Resemblance is not to be analyzed in terms of identity. Hence same­
ness with respect to token is not the same as (is not identical with) sameness
with respect to type. The word “same” is fundamentally ambiguous.
13)This is not a view which 1 accept. But it is an attempt to grapple with the

problem.
14Again, it may be held that sameness of token and sameness of type is sameness
 442 PACIFIC PHILOSOPHICAL QUARTERLY

in exactly the same sense, viz. identity. This Realist view seems to be nearer the
truth of the matter. I think it is a bit crude as it stands, because it appears to
require recognition of a universal wherever we recognize sameness of type, a
universal corresponding to each general word. However, the rightness or wrong­
ness of the answer is not what is in debate here. The point is that the philosophical
problem of the nature of sameness of type is faced, not evaded.
15
By comparison, what does Quine offer us? He simply says:

16)
That the houses and roses and sunsets are all of them red may be taken as ultimate and
irreducible, . . . (p. 1 0 .)

17
What does he mean by this? This remark might be made by a Realist, or at
any rate by a Realist who believes that redness is a property. But, of course,
Quine is engaged in rejecting Realism, personified by the unfortunate McX.
18)
It is natural to interpret him instead as saying that, although these tokens are
all of the same type, yet we have no need to consider what sameness of type
is. (And, a fortiori, sameness of type is not a matter of identity of property.)
If this is the way to interpret Quine, then is he not an ostrich about types?
Like an Oxford philosopher of yore, he keeps on saying that he does not deny
that many different objects are all of them red, but what this ostensible sameness
is he refuses to explain (except to say it is ultimate and irreducible). Instead,
he thrusts his head back into his desert landscape.
19) But perhaps there is a still deeper level of scepticism in Quine. Perhaps he

would object to this foisting upon him of talk about types. Suppose a is red and
b is red, then, Quine might say, we can by a convenient abbreviation say that
a and b are both red. If a is red and b is red and c is red, we can by a convenient
abbreviation say that a, b and c are all of them red. But nothing here justifies
talking of sameness of type, unless this too is mere abbreviation.
20) Such scepticism cannot be maintained. It is true that 'a and b are both red’

is an abbreviation of 'a is red and b is red.’ But the abbreviation does not hold
just for these particular sentences (much less for the above sentence-tokens), but
is a rule-governed, projectible, transformation which we are capable of applying
to an indefinite multiplicity of sentences. And what is the rule? It goes something
like this. Suppose that we are given sentences of the form ‘a i s __ and b i s ___
.’ If but only if the two blanks are filled by the same predicate, it is permitted
to rewrite the sentence as 'a and b are both__ ,’ with that same predicate in the
new blank. But ‘same predicate’ here is a type-notion. It is not meant that the
very same predicate-token be plugged successively into the three gaps!
21)
It appears, then, that just to understand phrases like ‘are both red’ requires
that we understand at least what a predicate-type is. And if this notion is
understood, and at least at a Moorean level accepted, then there can be no bar
to understanding, and at least at the Moorean level accepting, type-notions
generally. Some account must then be given, reductive or otherwise, of what
sameness of type is.
22) But perhaps Quine failed to appreciate this point when he wrote “On What

There Is.” The insight on which the argument of the penultimate paragraph is
based was not available to contemporary philosophers until the work of Donald
 AGAINST “ OSTRICH” NOMINALISM 443
cf. Davidson 1965
Davidson. For this, see Davidson, 1965, who criticizes Quine for a similar
failure to appreciate the projectible semantic structure of sentences attributing
beliefs in Word and Object (1960).
23)It may be, then, that Quine did not perceive at least the full urgency of the

need to give an account of types. But however it was with Quine, (or is with
Devitt), the distinction between tokens and types cannot be ignored. Hence a
philosophical account of a general sort is required of what it is for different
tokens to be of the same type. To refuse to give such an account is to be a
metaphysical ostrich.

II. Quine’s criterion of ontological commitment

24) But there is, of course, something else which insulates Quine from the full
impact of the problem of types, from the problem of the One over Many. The recusar
que
insulating material is his extraordinary doctrine that predicates involve no on­ certos
tokens
tological commitment. In a statement of the form ‘Fa,’ he holds, the predicate possuem
um
‘F’ need not be taken with ontological seriousness. Quine gives the predicate type,me
what has been said to be the privilege of the harlot: power without responsibility. faz um
avestruz
O que esta The predicate is informative, it makes a vital contribution to telling us what is metafísic
passagem o
quer dizer the case, the world is different if it is different, yet ontologically it is supposed
exatamente?
not to commit us. Nice work: if you can get it.
25) It is at this very point, however, that Quine may protest, as Devitt does on

his behalf, that his Nominalism is at least not an Ostrich Nominalism. For
although Quine is perfectly cavalier about predicates, he is deadly serious about
referring expressions. Suppose that a statement meets three conditions. (1) It
makes ostensible reference to universal. (2) We account it true. (3) It is im­
possible to find a satisfactory paraphrase of the statement in which this reference
to universal is eliminated. Under these conditions, Quine allows, indeed insists,
we ought to admit universal into our ontology. Perhaps the three conditions
cannot be met, but if they can be met, why then Quine will turn Realist.
26) 1 grant freely that to put forward such a set of conditions is not the behaviour

of a philosophical ostrich. On the other hand, I do think that Quine is an ostrich

with respect to the One over Many argument. Furthermore I think that Quine
(and his followers) have been distinctly perfunctory in considering the many
statements which answer to conditions (1) and (2) and which appear to answer
to condition (3).
27)In Chapter Six of my book^ I consider the statements:

(1) Red resembles orange more than it resembles blue

(2) Red is a colour
(3) He has the same virtues as his father.

Basing myself upon work by Pap (1959) and Jackson (1977), I argue that these
statements cannot be analyzed in a way which removes their ostensible reference
to universal, or at least to property-instances. (I try to show the incoherence
of the doctrine of property-instances, that is, particularized properties, in Chapter
Eight.)
o que é ?
 444 PACIFIC PHILOSOPHICAL QUARTERLY

28)It would in fact have been desirable also to have made reference to Hilary

Putnam’s “On Properties” (1970) which considers the statement:

(4) There are undiscovered fundamental physical properties.

To this might be added an example suggested by David Stove:

(5) Acquired characteristics are never inherited,

and many others.

29) Now we might expect reasonably extended discussions of examples of this
sort in Quine. Our expectation, however, is disappointed. In “On What There
Is” he does mention:

(6 ) Some zoological species are cross-fertile

and says that, unless we can paraphrase it in some way, it commits us to

“abstract”^ objects, viz. species. But he does not say what account he would
give of (6).
30) As Devitt points out, in Word and Object (1960, §25), Quine does give brief

consideration to:

(7) Humility is a virtue

along with:

(8 ) Redness is a sign of ripeness.

For (8) he suggests

(8 ') Red fruits are ripe

which perhaps may be allowed to pass. But (7), which resembles (2), cannot
be rendered, as he seems to suggest, by:

(7') Humble persons are virtuous.

(31)
First, the truth of (7) is compatible with there being humble persons who are
not virtuous. Indeed, it is compatible with no humble persons being virtuous.
For it may be that every humble person is so full of glaring faults that, although
they have the virtue of humility, they are not virtuous persons.
(32)Second, and more seriously, the truth of (7') is compatible with humility not

being a virtue. Consider an example suggested by Graham Nerlich. Suppose it

was true, and well known to be true, that tall people are always virtuous.

(7") Tall persons are virtuous

is exactly parallel to:

 AGAINST “ OSTRICH” NOMINALISM 445

(7') Humble persons are virtuous.

But nobody would wish to suggest that it would then be a truth that:

(7'") Tallness is a virtue.

So not only does (7) fail to entail (7') which was the first objection, but (7')
fails to entail (7).
33)As Devitt says, Quine then postpones general discussion of the problem of

“abstract objects” until Chapter VII. In that chapter, Quine, without discussing
examples, suggests that all apparent reference to attributes and relations should
be dispensed with in favour of talk of “eternal” open sentences (or general
terms) and/or talk of classes.
34) Here, I agree, he has moved beyond his original position to some form of

Predicate and/or Class Nominalism. But he does not discuss the rather well-
known difficulties for these varieties of Nominalism. (Devitt, it may be noticed,
appears to think that the difficulties are insoluble.)
(35)
It seems, then, that Quine is in trouble, even under his own rules. But the
more important question, I think, is why we should grant him his rules. Devitt
can only say that:

. . . we need a criterion of ontological commitment. Perhaps Quine’s criterion has dif­

ficulties, but something along that line is mandatory.

After this less than full-blooded defence, one can only ask ‘Why not a criterion
which allows predicates a role in ontological commitment?’
(36) At this point, appeal may be made to semantics. Devitt makes such an appeal.

He says that one can give the truth-conditions of ‘Fa’ by saying that it is true
if and only if ‘a’ denotes some particular which ‘F’ applies to. He says that this
shows that ‘Fa’ can be true even though the ‘F’ carries no ontological commit­
ment. But two points may be made in reply. First, there may be alternative, and
perhaps more satisfying, ways of giving the semantics for ‘Fa.’ Devitt offers no
argument against this possibility. Second, and more important, the semantics
of ‘applies’ has been left totally obscure. The Realist may well argue, correctly
I believe, that a convincing account of the semantics of ‘applies’ cannot be given
without appeal to the properties and/or relations of the object a. (I owe this point
to John Bishop.)

III. Problems for Realism

(37)

Besides supporting Quine in his rejection of the One over Many argument, O
Devitt also argues directly against Realism. He confines himself to the problem, problema
central
familiar to all Realists, of how particulars stand to universals. I agree with Devitt para o
realista é:
that this is the central difficulty in the Realist position. So 1 will finish what I como o
particular
have to say by making some remarks about it. But Devitt’s own remarks are se
relaciona
brief, and I think it best to expound the problem anew. com o
38) The problem is a sub-problem of the problem about the nature of particulars. universal
 446 PACIFIC PHILOSOPHICAL QUARTERLY

For one who denies the existence of properties in re (whether these properties
be universals or particulars), particulars are a sort of structureless blob. They
can have parts. Predicates can be hung on them, concepts applied to them, they
can be herded into classes, they may even have resemblances to other particulars
if a Resemblance Nominalism is adopted, but they lack real internal structure.
For those who accept properties in re, however, particulars are sort of layer-cake.
The one particular somehow unites within itself many different properties (an­
other One over Many). The question is: how is this possible?
(39) The problem divides at this point because a defender of properties in re may

develop the theory of particulars in different ways. According to one view, a

particular is nothing but its properties. It is not, of course, a mere class of
properties, but is a certain bundling of properties. A certain relation holds
between all and only the properties of a particular, and the holding of this relation
is what makes it a particular. This “Bundle” view in turn divides into two,
because the properties in the bundle may be conceived either as universals or
as property-instances. Russell held the Bundle view in its first form, Donald
Williams in its second.
40)However, the more orthodox view among those wh accept properties in re is

that, besides their properties, particulars involve a factor of particularity, an

individuating component. This view in turn divides into two in the same way
as before. There are those who make the properties into particulars. Locke is
a probable example. However, the more orthodox version of this more orthodox
view takes the properties to be universals.
41)
Since our special concern here is with the problems of Realism, we may
ignore the views which give an account of particulars by appealing to property-
instances. The view that a particular is nothing but a bundle of universals is
exposed to many grave difficulties (some of which I try to spell out in Chapter
Nine of my book), but 1 do not think that the difficulty raised by Devitt is among
them. The problem proposed by Devitt only arises, I think, if one holds (as 1
do hold) that a particular involves a factor of particularity (haeccitas, thisness)
together with properties which are universals. The question is then this: how are
the two components of a particular to be put together?
42 There are, broadly, two sorts of answer to the question which Realists have

given. According to the first, the factor of particularity stands in a certain relation
to the properties. It really is correct to speak of the related components of a
full-blooded particular. For this line of thought it is quite natural to reify the
factor of particularity and to think of it as a “bare particular.” This line of
thought, it seems further, ought to be reasonably sympathetic to the idea that
bare particulars might exist without any properties, and properties might exist
which are not properties of any particular. For why should not the relation fail
to hold? A synthetic necessity could be postulated to ensure that the factors only
exist in relation, but it is hard to see the necessity for this necessity.
43 But whether or not bare particulars can exist apart from properties, or prop­

erties from bare particulars, difficulties arise for this conception of a particular.
Let the relation be I, a bare particular be B, and wholly distinct properties of
the particular be P', F'. . . . An ordinary particular containing B will then be
constituted by a conjunction of states of affairs 1(B,P), I(B,P') . . . etc. The
 AGAINST “ OSTRICH” NOMINALISM 447

difficulty then is that I is a relation and so, on this view, is a universal. As a

result, a new relation of instantiation will be required to hold between I, on the
one hand, and the elements which it relates, on the other. The new relation will
then be involved in the same difficulty. The difficulty has been appreciated at
least since the work of F. H. Bradley.
44)
Various shifts may be attempted in the face of this regress, for instance, it
may be suggested that the regress exists, but is not vicious. Without arguing the
matter here, I will just say that I do not think that this way out, or any other,
succeeds.
(45) In common with many other Realists, I therefore favour the view that, while

we can distinguish the particularity of a particular from its properties, never­

theless the two “factors” are too intimately together to speak of a relation
between them. The thisness and the nature are incapable of existing apart from
each other. Bare particulars and uninstantiated universal are vicious abstractions
(in the non-Quinean sense of “abstraction,” of course!) from what may be called
states of affairs: this-of-a-certain-nature. The thisness and the nature are therefore
not related.
(46)
Frege says of his concepts that they are “unsaturated.” Fregean concepts are
not something mental. They are close to being the Realist’s properties and
relations. His idea, I think, was that the concepts have, as it were, a gap in their
being, a gap which must be filled by particulars. If we think of the particularity
of particulars as also “unsaturated,” then I think Frege’s metaphor is helpful.
47)All this is profoundly puzzling. As a result, Devitt is able to claim, not

implausibly, that all I have done is to substitute inexplicable mystery for the
relational view. Realism requires a relation between particularity and universality.
Yet to postulate such a relation appears to lead to insoluble problems. So, he
says, I simply “unite” the two factors in an incomprehensible manner.
48 I accept some of the force of this. But I have three things to say which I think

ought to make Devitt look upon this “Non-Relational Immanent Realism” with
a little more sympathy.
49 First, as was made clear already, the problem arises not simply where a

particular has a property, but where two or more particulars are related. Suppose
a has R to ¿7. If R is a universal, and a and b are particulars, and if we think
that a relation is needed to link a universal to its particulars, then we shall require
a further relation or relations to link R to a and b. This seems intolerable. It
seems much better, therefore, to say that, while we can distinguish the relation
from the particulars, yet the three “entities” are together in a way which does
not require any further relation to get them together. Now, if we think this way
about the polyadic case, it seems to me that when we go back to the monadic
case we ought in consistency take the same line, and deny that the particularity
of a particular is related to the properties of the particular. Contrariwise, if we
admit a relation in the monadic case, should we not admit an extra relation in
the polyadic case?
I hope that this generalization of the problem will at least show Devitt how
strong an intellectual pressure there is for a Realist to adopt a non-relational
view. It may be crooked, but it looks to be the best (Realist) game in town.
Second, 1 appeal to what Devitt says himself. He says:
448 PACIFIC PHILOSOPHICAL QUARTERLY

Talk of “particulars” and “universals” clutters the landscape without adding to our
understanding. We should rest with the basic fact that a is F.

Now, of course, I accept the second sentence just as much as Devitt. (There is,
as it w ere,/, a. in my philosophy as much as there is in Devitt’s.) Let us consider
the sentence. Devitt will surely admit that ‘a ’ is a token-word, picking out just
this thing a, while ‘F’ is a type-word, applicable, potentially at least, to many
things. Now why should we need two words of just this semantic sort to record
the basic fact? Does not some explanation seem called for? Is it so very extreme
an hypothesis that, while ‘a’ names a particular, ‘F’ captures something re­
peatable, something universal, about the situation?
I mdght add that I think that the dispute between Devitt and myself here is
an instance of a very deep dispute indeed. There are those who, apparently like
Devitt, think of reality as made up of things. There are others who, like me,
think of it as made up of facts or states of affairs. We cannot expect any easy
resolution of such an argument. (All the more reason to try to argue it of course.)
Third, I offer a second ad hominem criticism of Devitt’s position. Devitt
rejects the “One over Many.” But he agrees that the problems posed for Quine
by the arguments of Arthur Pap and Frank Jackson, retailed by me in the chapter
“Arguments for Realism,” are hard to solve. He makes no attempt to improve
upon the unsatisfactory paraphrases suggested by Quine of statements ostensibly
referring to universals. So it seems that he thinks that it may be necessary to
postulate universals. If he does have to postulate them, how will he solve the
problem of how universals stand to their particulars? I think he will end up
saying something similar to what I (and indefinitely many other Realists) have
had to say.

University of Sydney
Sydney, Australia

NOTES

'Devitt, following E. Prior, suggests that for the variable ‘X ’ be substituted the name ‘Armstrong.’
However, Devitt and Prior overlook the fact that ‘McArmstrong’ is ill-formed. ‘Armstrong’ is a
Lowland Scottish name.
^Devitt correctly noted that it was misleading to call the chapter “Arguments for Realism’’ in spite
of the fact that what I take to be the main argument for Realism, the One over Many, is deployed
in earlier chapters and is not deployed in Chapter Six. As Frank Jackson has pointed out, the title
should really be “ Arguments for Realism that work even if Quine is right about ontological com ­
mitment.’’
’Quine appears to mean by an “abstract” entity one that is outside space and time. This is a misuse
of the term, on a par with using “disinterested” to mean the same as “uninterested.” An abstract
object is one which can be considered apart from something else, but cannot exist apart from that
thing. Being outside space and time has no special connection with abstraction. He holds that biith
classes, if they exist, and universals, if they exist, are abstract in his sense. He also says that classes
are universals (1953, pp. 115-123), probably because he takes “universal” to be a convenient
synonym for “ abstract.” In fact classes are particulars, even if, as Quine claims, non-spatio-temporal
particulars. This is because, unlike universals, they are not ones which may run through many. There
 AGAINST “ OSTRICH” NOMINALISM 449

can be many instances of redness, but not many instances of the class o f men or the class of colours.
A “Nominalist,” for Quine, is simply one who does not recognize abstract objects in his sense, a
“Platonist” is one who does recognize them. So when he reluctantly admitted classes Quine became
a “ Platonist.” The misuse of all these terms has contributed to muddling a whole philosophical
generation about the Problem of U niversal.
‘‘Equally, supposing it to be true that:

(7"") Humble persons are amphibious

it does not follow that:

(7""') Humility is an amphibian.

REFERENCES

Armstrong, D. M. (1978): Nominalism and Realism (Volume 1 of Universals and Scientific Realism),
Cambridge University Press.
Davidson, D. (1965): “Theories o f Meaning and Leamable Language,” in Logic, Methodology and
Philosophy of Science, Proceedings of the 1964 International Congress, ed. Bar-Hillel, Y., North
Holland.
Jackson, F. C. (1977): “ Statements about Universals,” Mind, 86.
Pap, A. (1959): “ Nominalism, Empiricism and Universals: 1,” Philosophical Quarterly, 9.
Quine, W. V. O. (1953): From a Logical Point of View, Harper and Row.
Quine, W. V. O. (1960): Word and Object, M.I.T. Press