(Please, request original PDF from the author)

The Platonism vs. Nominalism Debate from a Meta-metaphysical Perspective1

Guido Imaguire
Universidade Federal do Rio de Janeiro (UFRJ)
[email protected]

Abstract: in this paper I present 5 different approaches to the debate between

Platonism and Nominalism: (1) the quantifier approach, (2) the reductionist approach,
(3) the mind/language dependence approach, (4) the extension versus intension
approach and (5) the hierarchichal approach. It one has its advantages and
disavantages that have to be discussed in detail.

Keywords: Platonism, nominalism, metametaphysics, existence.

The debate between Platonism and nominalism (here: “PN debate”) is persistent, ubiquitous
and multiform. Over the centuries, it has been present in different philosophical disciplines
and assumes, even in the same discipline, a variety of forms. The great variety of formulations
makes it difficult to understand what exactly is at stake in the PN debate.

Another problem that makes the PN debate particularly challenging is that, very often, the
choice between the two positions seems to be less rigorously theoretical and more
“ideological”, at times almost aesthetic. Szabó2 e.g. characterizes the dispute in terms of a
difference in “temperament”. Indeed, some nominalists express antipathy for a certain class of
entia non grata like universals or abstracta, without making explicit the reasons for their
dissatisfaction. The ideological nature is also clear from the adjectives used to describe the
disputed entities: they are “extravagant” or “repugnant”3. Unfortunately, not every
philosopher engaged in this debate is honest the way Goodman and Quine4 are in an opening
passage of their seminal paper on the issues: that abstract entities are to be renounced is a
“philosophical intuition that cannot be justified by appeal to anything more fundamental”.

1
My thanks go to Valdetonio Pereira de Alencar, Thiago Xavier de Mello and to an anonymous referee for many
comments and criticism.
2
SZABÓ, Zoltan – “Nominalism”. In: LOUX, Michael and ZIMMERMAN, Dean (eds.) – The Oxford
Handbook of Metaphysics. Oxford: Oxford University Press, 2005.
3
See LOWE, Jonathan – A Survey of Metaphysics. Oxford: Oxford University Press, 2002.
4
GOODMAN, Nelson and QUINE, William van Orman – “Steps Toward a Constructive Nominalism”. In: The
Journal of Symbolic Logic. 12 (1947), p. 105–22.
1
Later, Quine5 declares his “aesthetic sense for desert landscapes”. Now, how should we
evaluate intuitions or aesthetic sense?

All this has to do with what we can call the “approach problem”. This problem could be
formulated as a simple question: what exactly is at stake when Platonists and nominalists
disagree over their basic metaphysical convictions? It seems to me that the exact nature of the
PN debate is not clear, even to the most engaged participants. Of course, it seems implausible
that for centuries the two parties have just been talking past each other. Even metaphysical
deflationists like Hirsch6 consider the dispute not merely “verbal”. But perhaps the problem is
not a single disputed issue, but rather a whole bundle of disagreements.

In this paper, I try to make explicit some intuitions underlying the PN debate. I offer a general
analysis, i.e. an analysis that is valid not just for a single discipline, such as ontology,
philosophy of language or philosophy of mathematics, but for many disciplines. I think it is
important to reflect on this metaphysical dispute from a meta-metaphysical point of view in
order to gain clarity about what exactly has moved generations of philosophers to champion
one side or the other with almost the passion of a deeply felt ideological commitment.

1. Standard Formulation: To be or not to be

The standard way of framing the PN debate is in terms of existence. The Platonist defends the
existence of certain “problematic” entities, while the nominalist rejects it. Of course, for the
Platonist these entia disputanda7 are not problematic at all. To the contrary, their ontological
status is even higher and their identity criteria better than the criteria for rival entities, i.e.
particular concrete objects. To call those entities “problematic” is simply a nominalist
prejudice – as far as the Platonist is concerned.

The obvious problem with the standard formulation is that in order to decide whether or not
the disputed entities exist, Platonists and nominalists must first agree on the criteria for
existence. Unfortunately, we have far too many different conceptions of existence; and
possibly worse, we have different concepts of existence.8 Thus, all the approaches I present in
5
QUINE, William van Orman – “On What There Is”. In: QUINE, William van Orman – From a Logical Point
of View. Second edition, revised, Cambridge, MA: Harvard University Press, 1964, p. 1–19.
6
HIRSCH, Eli – “Ontology and Alternative Languages”. In: CHALMERS, David and MANLEY, David and
WASSERMAN, Ryan (eds) – Metametaphysics Oxford: Oxford University Press, 2009, p. 252.
7
I will use the phrase entia disputanda to keep my characterization of the PN debate more general. Indeed, my
characterization of nominalism should be able to explain the motivation for rejecting both universals and
abstracta.
8
The distinction between “concept” and “conception” is usually made in theory or justice (see e.g. John Rawls
in his A Theory of Justice). A difference in the conception of justice is not to broad as a difference in the concept
2
the next sections are intended as explanations of what is meant in this debate – or should be
meant – by “existence”.

A methodological issue arises here: can we decide questions of existence without giving a
trivial answer? When the nominalist proposes a criterion that trivially excludes abstract
entities (e.g. to exist means to be located in space and time), the discussion is no longer
interesting. But, in the same way, the Platonist should not propose a criterion that trivially
implies the existence of everything (e.g. everything that can be the subject of a true
proposition has existence).9

In the rest of this paper, I will present and evaluate five approaches to understanding how
Platonists and nominalists conceive of existence. The five approaches are:

(1) the quantifier approach

(2) the reductionist approach
(3) the mind/language (in)dependence approach
(4) the extension vs. intension approach
(5) the hierarchical approach

The first three approaches are certainly well known strategies, but we shall see that they are
not fully adequate for understanding the PN debate. Indeed, it seems to me that approaches (4)
and (5) are more satisfactory and so can bring new and better insights to this old debate. I do
not claim that the approaches on this list are exhaustive or pairwise exclusive. Some are
connected, perhaps even overlapping. But I think that just making the distinctions will help us
to clarify the issue.

Two preliminary observations are important here. First, note that there is an important
asymmetry between Platonism and nominalism regarding the general nature of the debate. No
Platonist would ever reject the view that Platonism is a metaphysical position, while the
nominalist can either accept nominalism as a metaphysical position or simply reject the whole

of justice. In the case of the notion of existence: two metaphysician can agree about the concept of existence, but
differ about its conception. So, virtually any one will agree that an existing entity has a stronger ontological
status than a non-existing entity. But what is meant by existence remains vague.
9
I am not suggesting that Meinong (or Neo-Meinonguians) trivializes the dispute in a Platonist way. His
generous criterion for being an object is not a criterion for existence: subsisting and non-subsisting objects are
non-existent. But, in a sense, the dispute becomes pointless in his hands: if I correctly understand his position,
concrete entities trivially (per definition) exist, and abstract ones trivially (per definition) subsist (and do not
exist).
3
enterprise of metaphysics. Because of this asymmetry, Carnap considered the Vienna Circle’s
position to be closer to that of nominalism:

It is therefore not correct to classify the members of the Vienna Circle as nominalists,
as is sometimes done. However, if we look at the basic anti-metaphysical and pro-
scientific attitude of most nominalists (and the same holds for many materialists and
realists in the modern sense), disregarding their occasional pseudo-theoretical
formulation, then it is, of course, true to say that the Vienna Circle was much closer to
those philosophers than to their opponents.10

Thus, the nominalist begins the competition with a kind of combinatorial advantage: while the
Platonist has just one option – he claims that P (“abstract entities exist” or something similar)
is true – the nominalist has two options – he can maintain that P is false or that it lacks any
truth-value.

Second, we should distinguish between nominalist claims and nominalist attitudes. Someone
can have a nominalist attitude and so aspire to eliminate the entia disputanda, despite
conceding that they “unfortunately” do exist, since he is unable to eliminate them. Russell and
Quine are good examples of this position. Russell defended the position that we should apply
Ockham’s razor whenever we can, but considered it impossible to eliminate all universals, in
particular relations. And Quine, for his part, would certainly have been happy to eliminate
numbers, but because of his own criterion, he was unable to do so. Many of the things stated
in the next sections are probably better suited for expressing nominalist attitudes than for
supporting nominalist claims.

2. Quantifier-approach

Given the risk that the whole debate between Platonists and nominalists could become
pointless, it was easy to convince both parties to accept an allegedly impartial criterion for
existence. Quine11 famously suggested that we should understand existence simply by means
of logical quantification. Quine’s criterion for existence was very simple: we should translate
the true sentences of our best theories into canonical logic and see what the bound variable
must range over for the result to be true. The elements of the domain of quantification are the
ontological posits. Ideological impartiality was assured by the inputs of outsiders: not the

10
CARNAP, Rudolf – “Empiricism, semantics, and ontology”. In: CARNAP, Rudolf – Meaning and Necessity:
a study in semantics and modal logic. Chicago: University of Chicago Press, 2nd edition, 1956, p.215. See
further HAHN, Hans – “Überflüssige Wesenheiten. (Occams Rasiermesser)”. In: SCHLEICHERT, Hubert –
Logischer Empirismus – Der Wiener Kreis. Munich: Wilhelm Fink Verlag, 1975.
11
QUINE, William van Orman – “On What There Is”. In: QUINE, William van Orman – From a Logical Point
of View. Second edition, revised, Cambridge, MA: Harvard University Press, 1964, p. 1–19.
4
philosopher, but the scientist determines our best theories and thus decides which sentences
we should consider true. The philosopher’s task consists only in translating these sentences
into canonical logic. In some cases, he should also make some paraphrase to eliminate merely
apparent commitments. In this way, Quine created new rules for managing the PN debate.

Quine’s suggestion seemed impartial, because he was forced by his own criterion to accept
Platonism in mathematics, despite his sympathies for nominalism. Thus, philosophers finally
saw a chance to make some objective progress in deciding questions of existence. Indeed,
only the unshakable conviction that Quine’s strategy for deciding questions of existence is
completely adequate could motivate anyone to undertake the vast theoretical enterprise of
reconstructing whole domains of natural science without using numbers, as Harry Field12
famously did. Many metaphysicians engaged in the Quinean game of formulating paraphrases
in canonical logic to make explicit the ontological commitments of some true sentences.

But after a period of general acceptance, contemporary metaphysicians became more and
more suspicious of Quine’s strategy. One early critical objection was already raised by
Alston13: Take the sentence S, which commits us to the existence of an entity e, and the
paraphrase S*, which does not commit us to e. Why should we suppose that it is S and not S*
that deceives us? Why not say that S* commits us despite its appearance? Why should we
suppose that the non-committing sentence is ontologically more adequate?

David Armstrong14 initiated a very influential line of criticism. In assigning the whole force of
the ontological commitment of a statement like ∃x(Fx) to the variable bound to the ∃, Quine
neglected the task of explaining the issue raised by the reference of the predicate term “F”.
Indeed, as Armstrong correctly reminds us, the traditional dispute concerning universals
emerged from the need to explain particulars’ sameness of nature: that both a and b are F’s –
a fact that Quine simply takes to be “ultimate and irreducible” and, therefore, in no need of
explanation. Simply supposing an additional abstract entity, the universal Redness, and
explaining the fact that a and b are both red by virtue of participating in it, as the Platonist
does, seems to be no theoretical improvement at all. Whether or not this is true is an
interesting question to investigate, but to decide this, one must abandon our comfortable meta-
ontological perspective.

12
FIELD, Hartry – Science Without Numbers: A Defense of Nominalism. Princeton: Princeton University Press,
1980.
13
ALSTON, William – “Ontological Commitments”. In: Philosophical Studies 9, 1-2 (1958), p. 8-17
14
ARMSTRONG, David Malet – Universals and Scientific Realism, vols. I and II. Cambridge: Cambridge
University Press, 1978
5
Mellor and Oliver15 also gave some reasons for rejecting Quine’s criterion. First, the
availability (or non-availability) of a paraphrase is always restricted to a very limited
assemblage of examples. Second, the idea that we can derive ontological commitment from
language is in itself suspicious. For example, instead of supposing (like Quine) that the truth
of “a is F” commits us to the existence of a and not of F-ness, one could (like Frege) take
“falls under” to be a primitive relation and assert that “‘a is F’ is true iff there is a ϕ such that
‘is F’ designates ϕ and ‘a’ falls under ϕ” and derive the commitment to F-ness.

Szabó16 has pointed out another problem with this quantificational approach: it is not really
self-evident, as Quine proposed, that ∃ captures the ordinary existential idiom. Native
speakers of English would find it unnatural to derive from “there is a good chance that the
Supreme Court won’t choose a president again” the conclusion that “there is something that is
a good chance that the Supreme Court won’t choose a president again”. Genuine ontological
commitment seems to be more complex than a simple quantificational interpretation suggests.
Quine would probably be unimpressed by this argument; he could reply that exactly because
of this unnaturalness, we should search for a better paraphrase that makes our ontological
commitments more explicit – and obviously we are reluctant to accept the existence of an
entity like “a good chance”.

But there are more serious difficulties with Quine’s quantificational approach. Kit Fine17
showed (2009: 165-66) that contemporary “Quinean oriented” ontology has probably been
dominated and vitiated by a failure to recognize the most elementary logical form of existence
claims. It seems obvious that the ontological commitment to the existence of, say, integers is
stronger than the mere commitment to the existence of natural numbers, but in the standard
quantificational form the second – ∃x (x is an integer ∧ x is non-negative) – is stronger than
the first – (∃x (x is an integer). Furthermore, when the realist asserts the existence of entities
of a certain kind F, it seems very natural to suppose that he does not claim that “there is at
least one F”, as the quantificational form suggests, but rather that all F’s exist. A realist about
natural numbers, for example, is not saying that “at least one” natural number exists, but that
they all do. Thus, for Fine assertions of existence can be articulated in a more suitable way if
we accept existence not as a quantifier but as a predicate, similar to the predicate “is real” (for
all x, if x is a natural number, x is real), or even better, as a sentence operator (“it is really the
case that”). If this is correct, Quine’s quantifier conception is completely misguided. After all,

15
MELLOR, David Hugh and OLIVER, Alex – Properties. Oxford: Oxford University Press, 1997, p.13.
16
SZABÓ, Zoltan Gendler – “Nominalism”. In: LOUX, Michael and ZIMMERMAN, Dean (eds.) – The Oxford
Handbook of Metaphysics. Oxford: Oxford University Press, 2005.
17
FINE, Kit – “The Question of Ontology”. In: CHALMERS, David and MANLEY, David and WASSERMAN,
Ryan (eds) – Metametaphysics. Oxford: Oxford University Press, 2009, p. 165-166.
6
Fine notes that one can consistently accept that numbers are indispensable for the purposes of
science and claim, nonetheless, that a realm of numbers does not really exist “out there”.

A more general criticism of the existence-approach can be found in Schaffer18. According to

him, the contemporary focus on existential questions (conceived in terms of quantifiers or
not) is fundamentally misguided. Moreover, like Fine he thinks that ontological questions
cannot be reduced to questions of quantification, for ontological questions are substantial,
while quantificational questions are trivial. Inspired by a more Aristotelian tradition, Schaffer
suggests that ontology should answer the question of grounding: what is a ground for what?
This question does not generate a “flat” existent vs. non-existent ontology, but rather much
more interesting hierarchical structures. Of course, this suggestion is too general to orient our
choices and decisions. But Schaffer made the interesting point that existence questions in the
Quinean method presuppose grounding questions: from choosing the best theory to reading
the entity commitments, all the steps are interwoven with grounding questions. If this is
correct, the best approach to the Platonism-nominalism debate is the hierarchical approach
(we will come back to this).

Metaphysicians may adhere to the notion of existence, reality, or even grounding, but they all
agree on one point: some entities are ontologically “stronger” than others. But why do they
think so? In any case, there is no doubt that the traditional formulation of the PN debate in
terms of existence and that of the quantifier approach seem fairly compatible. But sometimes,
valuable pre-theoretical insights are hidden behind official statements. In the next sections, I
present alternative approaches to the debate in order to make explicit what is really meant
when the Platonist says that some entities exist, and the nominalist denies this. For each of
them, the existence approach is a kind of starting point.

3. Reductionist approach

The PN debate is also frequently formulated in terms of reductionism. According to the

reductionist approach, the acceptance or rejection of the precept “do not multiply entities
beyond necessity”, also known as “Ockham’s razor”, is the main point of divergence. In this
picture, the nominalist is an enthusiastic “shaver” who loves desert landscapes, and the
Platonist an unshaven, bearded hippie who prefers forests. In contemporary philosophy,
Russell was one of the most prominent advocates of reductionism in this sense. His theory of

18
SCHAFFER, Jonathan – “On What Grounds What”. In”: CHALMERS, David and MANLEY, David and
WASSERMAN, Ryan (eds) – Metametaphysics. Oxford: Oxford University Press, 2009, p. 356-362.

7
descriptions is usually considered a very effective strategy for eliminating not only Meinong’s
superfluous subsistent entities, but also many other undesired entities. Strickly speaking, this
view is more suggested than property defended: metaphers as that of Ockham’s razor shaving
Plato’s beard, and the nominalist living in the desert landscapes contributed to this common
view.19

This approach explains the nature of the debate in quantitative terms: the Platonist wants
more, the nominalist fewer kinds of entities. There is a great advantage here: the comparison
between them becomes very easy. We could even distinguish intermediate degrees of
nominalism and Platonism according to the number of kinds of entities theorists accept. The
link to the notion of existence seems to be direct: if we can reduce an entity (of kind) A to an
entity (of kind) B, then A is superfluous, i.e., strictly speaking, it does not really exist.

Although this is a widespread characterization of the PN debate, I consider it to be the worst

possible way of characterizing the debate. It fails to capture the very essence of the
divergence. The most obvious difficulty of this approach is that it does not do justice to
Platonist intuitions. The Platonist is certainly not a multiplicator entiae. Plato’s heaven is as
minimalist as Quine’s desert.

Let us first think about Platonism in the philosophy of mathematics. From an historical point
of view, there is no connection between reductionism and nominalism. Frege never intended
his project of reducing arithmetic to logic as nominalist in spirit. Reducing mathematics to
logic does not imply that mathematical entities are not existent or real. Also, from a
systematic point of view, there is no connection between nominalism and reductionism: When
the nominalist rejects the abstract entities of mathematics, and supposes, e.g., that the number
2 is something “in the heads” of mathematicians, or just some material inscription, he gets
instead of a single entity, an enormous multiplicity: there are many numbers 2. Who is
multiplying entities here, the nominalist or the Platonist? For, remember that for the radical
Platonist the concrete world of many material inscriptions is not really existing, it is just a
shadow of reality. Of course, the nominalist can insist that what matters is not the number of
entities, but the number of kinds. This is the moral of Lewis’s distinction between quantitative
and qualitative economy. By identifying numbers with brain processes or with ink marks, the
nominalist is reducing the realm of abstract entities to the realm of concrete particulars, and so
reducing the kinds of entities.

19
See also CHATEAUBRIAND, Oswaldo – Logical Forms. Part II: Logic, Language, and Knowledge. Coleção
CLE, Unicamp, for a defense of Realism and rejection of Ockham’s Razor.

8
But the quantity of kinds is not a good criterion for distinguishing Platonism from
nominalism. Take, e.g., the bundle theory, i.e., the ontological theory that reduces concrete
particulars to universals. Is it in spirit a Platonist or a nominalist theory? Bundle theory
accepts only one ontological category, and is nevertheless a radical form of Platonism. Trope
theory is a nominalist theory, and it has exactly the same number of kinds of entities as bundle
theory: just one. Actually, neither the Platonist nor the nominalist is interested in multiplying
entities or categories without necessity. Both the Platonist and the nominalist agree that in
ontology we must reduce. Indeed, establishing the inventory of the most fundamental
categories of all entities is a central task of any ontology, and we get the fundamental
categories by reducing the non-fundamental ones. Thus, ontology is always reductive.

Plato himself was not a defender of plurality, but of unity in plurality. Not the Many, but the
One was the object of Plato’s philosophy. The aim of his project was not to multiply entities
as much as possible, but the contrary: to overcome the apparent plurality of entities in the
empirical world by condensing them to unifying pure forms. And in most philosophical
theories, universal and abstract entities are introduced precisely to condense a plurality to a
unity: the universal instead of its multiple instances, word types instead of various tokens,
propositions instead of various synonymous sentences, the abstract number instead of many
brain processes, material inscriptions or equi-numerical sets, etc. Thus, neither the quantity of
categories nor the acceptance of reductive procedures, but the “direction” of the proposed
reduction (e.g. from tropes to universals or from universals to tropes) is what distinguishes
Platonism from nominalism. Thus, the reduction approach fails to capture what is really at
stake in the PN debate. Therefore, let us look for other alternatives.

4. Mind/language-dependence approach

The most famous slogan of medieval nominalism, from which this position gets its name, was
originally formulated in a linguistic fashion: universale est nomen or flatus vocis. But in this
formulation, the PN debate is not obviously a debate about what exists. For, when universals
literally are predicates, even the nominalist must concede that universals exist, for predicates
obviously exist. Thus, the point at stake here must be more subtle. Curiously, the
contemporary characterization of the predicate nominalism proposed by Armstrong20 is no
better than a caricature: that a is F in virtue of the predicate “F” being truly applied to a seems

20
ARMSTRONG, David Malet – Universals and Scientific Realism, vols. I and II. Cambridge: Cambridge
University Press, 1978, p.12-13.
9
hardly defensible.21 Indeed, one way of thinking about the PN debate could be based on the
relation between predicates and universals. There is a sense in which one could claim that
universals are dependent on predicates. And since predicates are mind-dependent, due to the
transitivity of the dependence relation, universals must also be mind-dependent.22 Let us
consider this in more detail.

I will not distinguish here between the mind dependence and the language dependence of
universals: Hobbes held a linguistic variant of nominalism (only words are universal), Locke
a mental variant (universals are mental representations). Of course, mind is not language, and
predicates are not concepts. But both variants agree in the basic ideology: there are no
universals “beyond” or independent of human representation. Thus, I will simply speak here
in terms of mind-dependence. The thesis that universals are concepts (i.e. mental entities) is
often called “conceptualism” and as such is distinguished from nominalism. But, in my
general characterization as an anti-Platonist position, I consider it a form of nominalism.

Mind-dependence is often – and I think correctly so – associated with arbitrariness (in the
sense of “non-objectivity”). To say that universals are predicates is just a rhetorical manner to
say that they are arbitrary, in the same sense in which the choice of the predicates of a given
language is often supposed to be. In this sense, when the nominalist claims that universals do
not “exist in their own right”, i.e., they do not exist prior to our language, he is simply
suggesting that the set of predicates, and thus the set of properties, which we use to describe
the world is arbitrary. Thus, there are no “elite” properties, i.e., properties that “carve reality
at the natural joints”. Some philosophers seem to approach the PN debate along these lines.

One first example is the introduction of Mellor and Oliver to their reader Properties:

The most important questions about the kind of sameness, difference and change that
properties embody concern their reality and objectivity. Do particulars change or stay the
same, resemble or differ from each other, independently of how we think of or describe them?
That is, do properties exist in their own right – and if so which?23

In this passage the reality of universals is directly associated with objectivity. The authors
thereby suggest a strong connection between the PN debate and another major philosophical
issue: the debate between realism and idealism.

21
Perhaps for some medieval nominalists like Roscelin and Ockham the particular members of species really had
nothing in common except the species-name that referred to each member – or, at least, they said things like this.
22
Indeed, the medieval PN debate emerged from discussions about Porphyry’s commentary on Aristotle in the
third century, concerning the question of whether or not genera and species are mind-dependent.
23
MELLOR, David Hugh and OLIVER, Alex – Properties. Oxford: Oxford University Press, 1997.
10
A similar line of reasoning can be found in Bambrough’s “Universals and Family
Resemblance”24. The author interprets Wittgenstein’s notion of family resemblance as a
solution to the problem of universals. Independently of whether his interpretation of
Wittgenstein is correct or not, the interesting point is that he clearly understands the debate in
these terms: while the Platonist assumes universals to justify objective classification, the
nominalist stresses that every classification is arbitrary and depends on human interests and
purposes. Thus, to say that universals exist is the same as to say that some predicates (those
expressing universals) have an objective basis, a fundamentum in re. And to say that
universals do not exist is the same as to say that our predicates have no objective basis and are
merely arbitrary.

The mind-dependence approach is not just present in the PN debate on universals. The debate
on abstracta also follows this line of reasoning. The conventional Platonist in the philosophy
of mathematics views the claim that mathematical entities like numbers exist independently of
us as an argument for the objectivity of mathematics. If abstract entities like numbers exist
“out there”, they must have objective features independently of us. For the nominalist the
option remains to give up the objectivity of mathematics or, perhaps better, to provide another
explanation for it. The current metaphor used to characterize both positions in the philosophy
of mathematics fits very well with the dependence approach: either we “discover” or we
“create” mathematical entities and truths.

There are other authors approaching the debate in these lines, we cannot present here. One last
interesting case is the interpretaion of Michael Dummett on the debate on the Theory of Truth
in his “Truth and Other Enigmas”25 and in “Logical Basis of Metaphysics”26. According to
him, Platonism is associated with the traditional notion of truth as correspondence, while
nominalism is normally commited to a constructive (mental or linguistic) notion of truth.27

But some care is needed here in order not to transform nominalism into a caricature. In
particular, I would not characterize nominalism, as did Bambrough, as claiming that every
classification is arbitrary. Armstrong28 introduced the important distinction between

24
BAMBROUGH, Renford – “Universals and Family Resemblance”. In: The Proceedings of The Aristotelian
Society, 1960. Reprinted in LANDESMAN, Charles (ed.) – The Problem of Universals. London: Basic Books,
1971.
25
DUMMETT, Michael – Truth and Other Enigmas. Cambridge, MA: Harvard University Press, 1978.
26
DUMMETT, Michael – The Logical Basis of Metaphysics. Cambridge, MA: Harvard University Press.
27
I thank the anonymous reviewer for this indication.
28
ARMSTRONG, David Malet – Universals and Scientific Realism, vols. I and II. Cambridge: Cambridge
University Press, 1978.
11
subjective and objective forms of nominalism. Bambrough recognized only the subjective
form of nominalism. But many nominalists are pleased to say that, once we have selected the
predicates of our language, classification is objective. But this does not mean that in order to
justify objective classification we need to suppose the existence of abstract entities like
universals. Take one example in the spirit of the quoted passage from Mellor and Oliver:
some days ago this banana in my hand was (t1) green, but now (t2) it is yellow. Did it really
change its properties “independently of how we think of or describe” it? In everyday
language, we must say that the banana changed. It was green, and now it is no longer green, it
is now yellow. In Goodmanian-style language, with the predicates “greellow” (“green until t1
and yellow after t1) and “yelleen” (“yellow until t1 and green after t1”),29 we are led to a
different conclusion.

According to this language, there was no change at all: the banana was greellow at t1 and is
still greellow at t2. But the correct overall conclusion is not that the change of properties in the
world depends on language. The predicates “green”, “yellow”, “greellow” and “yelleen” are
not the reason for the distribution of properties in the world. Our object was objectively green
at t1, and yellow at t2, or, if you like, it was objectively greellow the whole time – just as a
(relatively) stable yellow object like the sun was objectively yellow the whole time. Thus, the
arbitrariness lies in the choice of the language we use to describe reality. But once we have
our predicates, description is fully objective.

Thus, not every nominalist is skeptical about objective property distributions in the world.
Quine, Goodman and many other nominalists would not deny the commonsense evidence that
things objectively change. Quine agrees with “pre-philosophical common sense”, i.e. that,
beyond language there are “red houses, red roses, red sunsets”. As he puts it: “that the houses
and roses and sunsets are all of them red may be taken as ultimate and irreducible. But, and
this is the point, they do not commit us to say that “there are” universals like Redness.30 The
problem is the “popular and misleading manner of speaking” to say that houses, roses, and
sunsets “have something in common” and consequently, that this additional thing (Redness)
exists. For the nominalist, the objectivity of property attribution does not imply the existence

29
In Goodman’s hypothesis, the predicates “grue” and “bleen” not only have parameters for time, but also
relative to observers “being observed before t and is green, or is not observed before t and is blue”. Thus, he is
not suggesting that things spontaneously change color. But this is not relevant for my point, in particular because
I chose a kind of thing that indeed (and regularly) changes color.
30
QUINE, William van Orman – “On What There Is”. In: QUINE, William van Orman – From a Logical Point
of View. Second edition, revised, Cambridge, MA: Harvard University Press, 1964, p. 1–19.
12
of universals. The reaction of a Platonist like Armstrong31 is simple: Quine is an ostrich,
because he does not take predicates with any ontological seriousness.

Because of all this, we should not claim that the dependence approach is the right way to
understand the PN debate, but only that it has been approached along these lines. This
approach makes the PN debate strongly epistemological, and many would reject this, in
particular Rodriguez-Pereyra32 and Crocket33. Nevertheless, to claim that the PN debate is an
ontological debate does not imply that it does not have any direct epistemological
consequences. Issues in metaphysics usually have consequences in all other fields of
philosophy. Thus, relativism is not a necessary consequence of nominalism, but the fact that
many nominalists have also defended some kind of relativism (e.g. Hume, Quine, Goodman)
seems to be more than a mere coincidence. If universals exist prior to language, there will be
an objective criterion for the distinction between more or less adequate languages (even if we
are not able decide this ranking).

Although essentially the dependence approach is formulated in terms of predicates and

concepts, similar reasoning concerning arbitrariness also applies to sets or classes. For the
Platonist, objective properties ground facts of similarity or nature-sameness: a and b are both
F. It seems very natural, even mandatory, to recognize that this cat is similar to that other cat,
but very different from that elephant. To be a cat and to be an elephant are objective
properties and, thus, they ground similarity and dissimilarity in reality. These objective
properties determine the classes or sets we take to be natural: the set of cats, the set of
elephants, and so on. Our practice of “building” natural sets of objects is, so to speak, “well-
founded”, so the Platonist.

Contrary to this, the nominalist thinks that we could arrive at a classification according to
which this cat and that elephant belong to the same set, while that other cat does not. There
are no elite sets, all arbitrary miscellaneous sets of particulars correspond to a property* (the
“*” should indicate that we are dealing here with an “abusive” notion of property). If we take
all the possibilia of this domain, we get what Lewis34 called “abundant properties”. And thus,
Lewis’s claim that any miscellaneous set of particulars is a property* is a strong form of

31
ARMSTRONG, David Malet – Universals and Scientific Realism, vols. I and II. Cambridge: Cambridge
University Press, 1978, p.16.
32
RODRIGUEZ-PEREYRA, Gonzalo – “What is The Problem of Universals?”. In: Mind 109 (2000), p 255-
273.
33
CROCKETT, Campbell – “The Confusion over Nominalism”. In: The Journal of Philosophy, vol. 47, 25
(1950), p. 752-758.
34
LEWIS, David – “New Work for a Theory of Universals”. In: Australasian Journal of Philosophy, 61 (1983),
p. 343–77.
13
nominalism. But, to accept that some sets are “more natural” than others, and that there are
even “perfect natural classes”, as Lewis did in the same paper, is certainly making a
concession to Platonism.

How can the Platonist react to the nominalist claim that every classification is man-made?
Reasonably, one good Platonist strategy for combating this thesis consists in recurring to the
authority of science to save objectivity. Some properties* correspond to genuine real
properties. Armstrong, Shoemaker and Mellor35 adhere to this strategy. Even some
nominalists are willing to accept that pure science should decide the question, either via
quantification (Quine) or via perfect natural classes (Lewis).

One advantage of conceiving the PN debate in these terms is that it enables us to distinguish
intermediate positions. Take, e.g., a world with n particulars and all corresponding abundant
properties, predicates and sets. Platonism can be defended in a radical or in a moderate form.
On the one hand, for the radical Platonist, if a is 3 years old and b is 3 years old, then one
must accept the existence of not only the predicate “to be 3 years old” and the set of things
that are 3 years old, but also of the universal to-be-3-years-old. Even the predicate “to be
identical with Churchill or with the Eiffel Tower” expresses a universal. This is an
excessively tolerant conception of universals. When we identify Frege’s concepts with
universals, he becomes a good example of this position, for every open sentence is an
unsaturated expression that corresponds to a genuine property. On the other hand, the
moderate Platonist finds it too permissive to accept that every predicate or every set
corresponds to a genuine universal. Thus, he usually proposes some restrictive criterion for
selecting the elite-predicates or elite-sets that correspond to genuine universals. This is the
position that Realism defends, e.g., cf. Armstrong, according to which we must select some
elite predicates based on a posteriori natural science.

The nominalist, for his part, can also assume two attitudes, a radical and a moderate one. The
moderate nominalist considers that some predicates, sets and properties are more natural than
others. Some predicates are “more natural”, others even “perfectly natural”. This is a kind of
nominalism, because “natural” is a primitive predicate, and the predicates do not denote
universals. This is the position of Quinton36 and Lewis37. The radical nominalist, on the other

35
ARMSTRONG, David Malet – Universals and Scientific Realism, vols. I and II. Cambridge: Cambridge
University Press, 1978. SHOMAKER, Sydney – Identity, Cause, and Mind: Philosophical Essays. Cambridge:
Cambridge University Press, 1984. MELLOR, David Hugh – “Properties and Predicates”. In: CAMPBELL,
Keith et allii. – Ontology, Causality, and Mind. Cambridge: Cambridge University Press, 1991.
36
QUINTON, Anthony – The Nature of Things. London: Routledge and Kegan Paul, 1973.
37
LEWIS, David – “New Work for a Theory of Universals”. In: Australasian Journal of Philosophy, 61 (1983),
p. 343–77.
14
side, thinks that all predicates, sets and properties are equal. When we assume some criterion
for selecting some putative elite-predicate, this criterion is probably only relative to our
arbitrary and/or pragmatic interest. Goodman is a good example of a nominalist with this
viewpoint.

Interestingly, the radical Platonist and the radical nominalist are very close to each other in
one respect. Both agree that all predicates are “on the same level”, i.e. there are no better
predicates. This coincides with Balaguer’s38 diagnosis concerning the debate in the
philosophy of mathematics. Balaguer noted that, surprisingly, radical Platonism and radical
Nominalism are much closer to each other than to intermediate positions. In mathematics, the
radical Platonist and the radical nominalist agree that all logically possible entities have the
same status (the Platonist thinks: they all exist, the nominalist: none of them exists). When we
exclude their statements about existence, the discourse of the radical Platonist and that of the
radical nominalist are the same. The same happens here concerning predicates: the radical
Platonist and the radical nominalist take all predicates to express properties of the same status:
the radical Platonist thinks they all express genuine properties, the radical nominalist thinks
they all express merely arbitrary properties.

5. Extension versus Intension Approach

There is widespread agreement that, in general, nominalists have sympathies for extensional
and Platonists for intensional entities. The terms “extension” and “intension” are used in
many different senses. I have in mind here not the semantic, but the ontological aspects of
extensional and intensional entities. The extension vs. intension distinction is not simply
black-and-white. Entities can be more or less extensional depending on their criteria for
individuation. The more detailed the identity criteria of an entity are, the more intensional
they are. Take sets and mereological sums, for example. Sets are extensional entities because
their identity criterion is supposed to be extensional – i.e. they are identified “exclusively” by
their members. But we can say that mereological fusions are even more extensional than sets.
Take two concrete particulars a and b. The sets {{a, b}, {a}} and {a, b} are different, while
the corresponding mereological sums ((a+b)+a) and (a+b) are identical. Mereological sums
are individuated exclusively by their ultimate constituents, while sets are not, they admit of
one additional “structural” element to distinguish them. By accepting only extensional
entities, we avoid additional merely formal entities. According to Goodman’s (1956) principle
of nominalism: if a and b are made up of the same constituents, then a=b. Take, e.g., a

38
BALAGUER, Mark – Platonism and Anti-Platonism in Mathematics. Oxford: Oxford University Press, 1998.
15
concrete pile of stones and their molecules. When you accept classes, you have two kinds of
concrete objects and (at least) two abstract ones (the set of stones and the set of molecules).
When you accept only mereological fusions, there is no additional entity: the pile of stones,
the molecules, the fusion of the stones and the fusion of the molecules are all just one single
entity. Because of this, a radical nominalist like Goodman preferred mereology to set theory.

When we see that fusions are more extensional than sets, we can understand Quine’s
statement that “… nominalism is distinct from the doctrine known in modern logic as
extensionalism”39. In fact, Quine understood “extensionalism” as the doctrine that properties
or qualities are to be replaced by classes, and since classes are abstract, the Quinean
nominalist has to reject extensionalism. But Quine would certainly agree that someone who
tries to reduce qualities to mereological sums (“red is the smallest particular that has all red
particulars as parts”) is absolutely loyal to nominalist ideology. Mereological nominalism is a
more radical form of nominalism than class nominalism: mereological nominalism rejects
both properties and abstractions, class nominalism rejects only properties.

Now, let us look for a more general characterization of the extension-intension distinction that
can be applied to many different entities. Take the following diagram:

i1, i2, i3…… ⇄ A

The basic interpretation of the diagram is just this: on the left side, we have a collection of
particular instances, all of them related by some relation of equivalence, and on the right side,
a corresponding general entity – the abstractum. The double arrow in the middle represents
the two possible transitions from one side to the other, i.e., something like abstraction or
participation. In this approach, the “ideological” difference between the Platonist and the
nominalist is a matter of which side is more fundamental, and how the other side is
“constituted” or “generated”. This diagram can be interpreted in different ways, and in each
interpretation, it reveals particular aspects of the PN debate.

The most obvious interpretation is by means of the particular vs. universal distinction. The
particular instances i1, i2, i3… share the common universal nature A; e.g., Socrates, Plato and
Aristotle are particulars (i1, i2, i3) that instantiate the property humanity (A). The Platonist
holds that the universal to be a man is prior to and independent of individuals; these
participate in the universal, but are not constitutive of it. The nominalist, in turn, considers the

39
Quine according to MANCOSU, Paolo – “Quine and Tarski on Nominalism” in Oxford Studies in
Metaphysics, Dean Zimmerman (ed), Oxford: Oxford University Press, 2008.
16
other way round to be more natural: prior are individuals, from whom, by means of the
equivalence relation of respect-similarity, we abstract humanity.40

The diagram can also be applied to the philosophy of language. On the first level, we have
many similar word tokens (i1, i2, i3) that correspond to a single word type (A). For the radical
nominalist, even on this very basic level we already have something, an abstract entity, in
need of explanation. Under the influence of Saarnio41, Quine42 recognized that, strictly
speaking, word types are universals: “they are not particular inscriptions, mounds of ink or
neon tubes, but general forms of inscriptions”. Thus, the radical nominalist must reduce word
types to similar classes of word tokens. The Platonist, from his generous point of view, cannot
see any philosophical problem at all on this basic level of word types. Particular inscriptions
are just instances of a general form. At the second level, supposing that word types are
somehow “generated”, a second process of abstraction takes place: different expressions,
words or sentences are entities (i1, i2, i3) that stand in the equivalence relation of being
synonymous to or having the same meaning (the abstractum A is here a word meaning, or a
proposition in the case of sentences). Thus, meaning is the new abstractum that the nominalist
has to explain away. Again, a typical nominalist prefers to eliminate abstract meanings from
his theory and explain language exclusively on the basis of words and concrete stimuli. The
typical Platonist, for his part, thinks of meanings as entities that exist prior to every language,
just as does Frege. Words are simply the labels of meanings.

The intensional-extensional approach fits very well with the usual identification of
nominalism with particularism. It also has the advantage of unifying the intuitions of different
kinds of nominalism. Of course, the rejection of abstract entities and the rejection of
universals are different things. Nevertheless, in this approach it becomes clear what is
common to both universals and abstract entities: they are intensional (an abstractum) in the
explained sense. The basic philosophical intuition becomes very simple: the nominalist is a
friend of concrete particulars and word tokens, the Platonist of universals, word types and
meanings. Indeed, even in the philosophy of mathematics we have a direct analogy: all
particular three-member sets (i1, i2, i3, …) are related by the equivalence relation of bi-
junction and correspond to the abstractum natural number 3 (A). Thus, Platonists are friends
40
It is generally assumed that on the left side, the domain of particular instances is somehow “chaotic”, and that
on the right side, there is a kind of organizational principle that establishes some intelligibility in the chaos. For
the traditionalist, asserting the instantiation of a property by a particular (or the falling of an object under a
concept) is the most basic form of establishing the intelligibility of reality. We can connect this with the
mind/language approach: the Platonist thinks that we discover intelligibility as immanent in reality, while the
nominalist thinks that we create or project intelligibility into reality.
41
SAARNIO, Uuno – Untersuchungen zur symbolischen Logik. Väitöskirja: Akateeminen Kirjakauppa, 1935.
42
MANCOSU, Paolo – “Quine and Tarski on Nominalism” in Oxford Studies in Metaphysics, Dean Zimmerman
(ed), Oxford: Oxford University Press, 2008.
17
of numbers, nominalists friends of sets (but, of course, even better friends of the concrete
elements of sets).

One issue that could be examined from this path is the epistemological aspects of the PN
debate. I think this approach makes it at least plausible that nominalists are typically
sympathetic to relativism, since extensions sub-determine intensions. Indeed, the transition
from extensions to intensions is always a logically suspicious procedure.

6. Hierarchical Approach

The last approach to the old problem allows, I hope, a fresh look. Actually, I don’t know any
explicit statement from metaphysicians that puts the debate in terms of hierarchies.
Nevertheless, this approach seems to be implicit in many central passages of different
important authors. According to it, the basic difference between the Platonist and the
nominalist is that they disagree in their intuitions about the hierarchical structure of reality.

A more obvious expression of this kind of intuition is the current definition of Platonism in
the philosophy of mathematics as the view that claims that mathematical facts and sentences
are “on a par with” empirical facts and sentences. This claim is often justified on the basis of
the so-called “Fregean argument”43. This argument is based on the principle that truth and
logical form have priority over the reference of the sub-sentential expression, i.e., when true
sentences have singular terms, these must refer44. Take the sentences “Socrates is wise” and
“2 is even”. Both sentences are true, therefore their subject terms must refer, and the referent
of “2” is a non-concrete entity. For the Platonist, there is no relevant ontological distinction
between these two kinds of sentences or these two kinds of facts. Both sentences are
composed of referential terms and qua true sentences have similar truthmakers. Something
similar is presupposed by the traditional abstract reference argument for universals: in
“wisdom is a virtue”, the term “wisdom” in the subject position must refer to a universal.

The nominalist, on the other hand, insists on the difference between both kinds of sentence,
denying that the second sentence has a truthmaker, or, if it does, then one of a very different
kind. For him, the first kind of sentence and fact is more robust or fundamental. Of course, “to
be on a par” is only a vague metaphor that helps to introduce the problem along general lines.
Without more explanations of what exactly is meant by “on a par with”, the Platonist insists

43
HALE, Bob – Abstract Objects. Oxford: Blackwell, 1987, p.11.
44
WRIGHT, Crispin – Frege's Conception of Numbers as Objects. Aberdeen: Aberdeen University Press, 1983,
p. 13-15.
18
on similarities, while the nominalist stresses differences, and the discussion runs the risk of
becoming pointless. At the end of the day, neither would the Platonist insist that both
sentences are absolutely similar, nor would the nominalist claim that they are absolutely
different.

To make some progress, the disputants must endow their respective favorite intuitions with
more substance. My proposal for understanding the PN debate in terms of hierarchies runs
like this: both the Platonist and the nominalist must admit some kind of ontological hierarchy.
At first glance, even the nominalist admits that “there are”, beyond concrete particulars, first
order properties, second order properties, and so on. When the nominalist claims that, strictly
speaking, only particulars really exist, he is not suggesting that Platonists have made a
categorization error. He is not suggesting that everything Platonist have classified as a
property, e.g. redness or wisdom, is better classified as a particular, like Socrates and this
table. Even the nominalist admits that redness and wisdom are properties – except that these
things do not really exist. The important difference between the Platonist and the nominalist
concerns the fixed or flexible position of entities in the hierarchy. While the Platonist claims
that an entity can “move” or “change its position” inside the hierarchy (or, at least, it can have
a kind of representative correlate on other levels), the nominalist insists that each entity has a
fixed position in the hierarchy. For the nominalist, the Platonist’s mistake is to treat properties
(first or higher order entities) as if they were objects (entities of the basic level) – objects of a
very special kind. Indeed, from the perspective of the nominalist, Plato’s original mistake was
to consider properties (such as to be red) as essentially like a special kind of object (redness).
Redness becomes a single thing (just as is Socrates). For the nominalist, reification is the
original sin of Platonism, and reification means here “to locate a higher order entity on the
fundamental level”. In other words, the problem is to consider properties as on the same level,
as “on a par with” particulars.

As we saw above, the Platonist insists that in “2 is even” and “wisdom is a virtue” the subject
terms “2” and “wisdom” refer in exactly the same way as “Socrates” in “Socrates is wise”.
Arguing along these lines, he concludes that 2 and wisdom exist in the same way that Socrates
exists, i.e. he takes second order properties (the number 2) and first order properties and puts
them on the same level as concrete particulars (like Socrates). For the nominalist, this leads to
great confusion. Ways are ways, and things are things. Wisdom is a way things are, and not a
special thing. The number 2 is the same way that all two-member collections are (or
something like that), and not something like the collected things.

19
A good example of this approach is found in Skyrm45. He characterizes Wittgenstein’s
Tractatus – I suppose correctly – as nominalist in spirit, because there are no higher order
facts. For example, that-wisdom-is-instantiated-by-Socrates is not a second order fact, but
“merely a re-description of the first order fact that Socrates is wise” (p. 202). More generally,
the tractarian nominalist denies “that there are distinct possible worlds which share all the
same first order facts” (ib. idem.). Higher order facts are only supervenient, or “ontological
free lunch”, or some other “thin” kind of thing. In any case, they are not of the same kind as
the “robust” first order facts. To make them as constitutive of reality as first order facts
involves an inversion of ontological hierarchies. Higher order facts are not on a par with first
order facts.

Take another example. As we saw above, Quine46 famously suggested that we should
understand existence simply by means of logical quantification. His criterion for existence
was, in principle, very simple: we should translate the true sentences of our best theories into
canonical logic and see what the bound variable must range over for the result to be true, and
the elements of the domain of quantification are the posits of our best theories. This
suggestion can also be seen as another example of reasoning in terms of hierarchies. We
usually quantify over objects, because objects are the “right thing” to quantify over.
Quantification over properties should be avoided, because properties are not objects and we
should not deal with properties as if they were objects. To quantify over properties is just like
making them into objects. If one assumes that predicates like “red” in “red is a color” do not
refer to a property, but rather to an abstract particular, one is flattening the hierarchy. If one
assumes that even in sentences like “red is a color” the term “red” refers to a property (and
not to a special kind of particular), one is preserving hierarchy and avoiding reification.

Against this approach to the PN debate, one might object that Frege, one of the most
prominent Platonists of the analytic tradition, argued for a rigid hierarchy. In his logical
hierarchy, objects are objects, first order concepts are first order concepts, and so on. This is
true, but it is not the whole truth. Indeed, for Frege an entity cannot change its position in the
hierarchy. But, and this is the point, he admitted a kind of representative correlate on different
levels. Horse is a concept, but the concept “horse” (see on the definite article) is an object.
Numbers first emerge in his system on the second level of predication, as second order
concepts, concepts of concepts. But, after this, he insists, “the” number 2 is an object just like
this table. Moreover, in his system sentences express propositions that can be true or false.

45
SKYRMS, Brian – “Tractarian Nominalism”. In: Philosophical Studies, 40 (1981), p. 199-206.
46
QUINE, William van Orman – “On What There Is”. In: QUINE, William van Orman – From a Logical Point
of View. Second edition, revised, Cambridge, MA: Harvard University Press, 1964, p. 1–19.
20
Thus, prima facie, truth values seem to be properties of sentences or propositions. But, as we
know, in his system The True and The False become objects (“of a very special kind”). Here
Frege is obviously reifying concepts, numbers and truth values – making them objects.

The nominalist has a good reason for being afraid about high levels entities being placed on
the botton of the hierarchy. Many restrictions have to be introduced in order to avoid logical
paradoxes. The plain nominalist lives a quite life only accepting particulars and irrestricted
quantification over them. The price the Platonist pays for permitting higher order entities
being placed at the first level is to introduce divisions between them. A comparison with
theories of justice is useful here. For egalitarian theory, everyone should get the same
treatment; for meritocratic theory, goods should be distributed depending on individual merit.
It is similar in metaphysics: Platonism is like egalitarianism, Nominalism is like elitism. For
the Platonist, all entities have similar ontological rights. For the nominalist, each “entity” (if
they are entities at all) has its proper place and its own merits; different caste should not to be
mixed.

Schaffer47 proposes that the most basic ontological question is not simply a question about
existence, but a question about ontological dependence: what are the grounds for what? If my
characterization of the hierarchical approach is correct, his suggestion is nominalist in spirit,
even if it seems to retain a Platonist import: questions of existence are positively trivial. Of
course, numbers and properties exist, he said. The interesting point is whether they are
fundamental or not. Indeed, our hierarchical nominalist can remain true to his metaphysical
convictions, even accepting that, in a non-rigorous sense of “existence”, everything exists.
But entities exist in different ways, some entities are basilar, more fundamental, while others
are derivative. And this is not the same as to say that these entities are not real: heat is not
unreal just because it is reducible to molecular movement.

On a more general level, we can challenge the claim that the hierarchical-approach is a
historically adequate interpretation of the old metaphysical debate. Overall, it probably would
make Plato a nominalist! Remember his pyramidal project, beginning with the world of
concrete particulars at the bottom, organized due to their participation in higher pure forms,
which are structured by formal notions, culminating in a notion with a higher degree of self-
identity that he founds in the principle of goodness. Of course, he introduced the “pernicious”
talk of different realms. But these realms are hierarchically separate; there are concrete
particulars at the very bottom and, overall, the inhabitants of higher levels are never “flat

47
SCHAFFER, Jonathan – “On What Grounds What”. In”: CHALMERS, David and MANLEY, David and
WASSERMAN, Ryan (eds) – Metametaphysics. Oxford: Oxford University Press, 2009, p. 356-362.
21
down” to lower levels. Perhaps there is a collapse in this approach similar to the one Balaguer
saw in the philosophy of mathematics and we saw in the language approach. And possibly the
disagreement is not about the task of building a hierarchical structure, but simply about
whether to build the structure from the bottom up or the top down. Thus, strictly speaking the
radical Platonist and the radical nominalist could agree if they were willing to bypass their
superficial formulations.

As long as we have a hierarchical structure somewhat like a skyscraper, and the nominalist
can live on the ground floor and the Platonist on the higher floors, they can live together in
peace and harmony. The problem only arises when the different levels are mixed together.
Thus, the radical Platonist who avoids mixing his beloved ethereal entities with the
nominalist’s grey concrete does not disturb the metaphysical peace.

22