One way to interpret the opposition between nominalists and realists in ontology is to trace it back to a disagreement in philosophical methodology: nominalists, as Quine famously stated it, “have a taste for desert landscapes” (Quine 1953), a certain “aesthetic sense”, which they will usually formulate as a principle of ontological economy, or “Ockham’s razor”. Realists, on the other hand, typically view themselves as starting their arguments with obvious facts, data of common sense, or “Moorean facts”. As Armstrong says, the fact that is central in establishing realism, namely “the fact of sameness of type, is a Moorean fact” (Armstrong 1980, 441), and in distinguishing it from “sameness of token”, “[philosophers] are only formalizing, making explicit, a distinction which ordinary language (and so, ordinary thought) perfectly recognizes” (ibid.). This is also how Quine understood the realist’s motivation; in “On What There Is”, Quine has the realist (McX) say the following: “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 I mean by the attribute of redness.” (Quine 1953, 9–10). And he concludes: “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.” (ibid.).

To sum up, there are (at least) two different methodologies one can adopt in metaphysics: if you adopt a principle of ontological economy (inspired by Ockham), all chances are that you will want to eliminate universals out of your landscape, and will end up a nominalist; if you adopt a principle of common sense (inspired by Moore), all chances are that you will end up a realist. Or so goes a traditional understanding of the debate.

In this paper, I want to challenge this traditional interpretation. First, I will try to show that realism is in fact motivated by a kind of principle of economy. In this, I share Imaguire’s conclusions who also challenged the widespread idea that a metaphilosophy of economy goes against realism (Imaguire 2015, sec. 3, 2018, sec. 7.2.2).Footnote 1 Furthermore, I will argue that a philosopher adopting a metaphilosophy of common sense should be a nominalist (more precisely an Ostrich nominalist).

To lay cards on the table, I personally adopt a metaphilosophy of common sense, and for that reason I also endorse (Ostrich) nominalism. I think there are good arguments in favour of a common sense metaphilosophy, but in this paper I will not try to develop these arguments. I will only try to establish the conditional thesis that if one is a common sense philosopher, then one should rather be a nominalist (and more precisely an Ostrich nominalist).

In the first section of the paper, I will present what I mean exactly by a “metaphilosophy of common sense”. In Section 2, I will compare this metaphilosophy of common sense with two versions of the principle of economy, one of which is compatible with the common sense methodology, while the other (radical economy) is at odds with the common sense methodology. In Section 3, I will present what I take to be the dialectical situation of the debate about universals. In Section 4, I will argue that the common sense methodology lays the burden of proof on the shoulders of the realist, not the nominalist. In Section 5, I will develop my central argument in favour of the view that the realist’s motivation (in favour of the existence of universals) lies in fact in some form of radical economy principle, i.e. a motivation to reduce the number of facts beyond what common sense prescribes. Therefore, a common sense philosopher should not accept the realist’s motivation. In Section 6, I will address some objections to this central argument.

1 The Methodology of Common Sense in Metaphysics

Different authors have meant different things with the phrase “metaphilosophy of Common Sense” (see in particular Double 1996; Boulter 2007).Footnote 2 For that reason, it will be important to give at least a summary overview of the kind of metaphilosophy I am relying on in this paper.

To cut a long story short, we can distinguish two versions of metaphilosophy of common sense, a strong version and a weak version.

The strong metaphilosophy of common sense says that common sense beliefs (or principles) are unassailable (or at least unassailable by philosophy – perhaps common sense could be overturned by science, but that is another issue). This strong version of common sense is what is referred to by both Richard DoubleFootnote 3 (who doesn’t endorse it) and Stephen BoulterFootnote 4 (who endorses it). We could formulate this metaphilosophy in the form of the following principle:

(Strong common sense) Never revise common sense: adapt your philosophical theories so that they never violate common sense.

The weak metaphilosophy of common sense, on the other hand, accepts the revisability of common sense beliefs (or principles), not only under the pressure of scientific evidence, but also (if the case happens) under the pressure of successful philosophical arguments. But it insists that, in the absence of such argumentative pressure, the common sense beliefs have a default justification and should be maintained (in other words, the “burden of proof” is always on the shoulders of anyone who wants to revise common sense beliefs). Why should we give a default justification to common sense beliefs (instead of others)? The main motivation here is that common sense beliefs are by definition the beliefs we start from, and for that reason do not need to be established in the first place (they are “already there”). This weaker principle can be formulated as follows:

(Weak common sense) common sense can be revised, but it is our starting point, and therefore has default justification until it is proven to be false.

Thomas Reid himself offered this kind of conception and defence of common sense, for instance in the following quote, concerning the common sense conception of free willFootnote 5:

“This natural conviction of our acting freely […] ought to throw the whole burden of proof upon that side; for, by this, the side of liberty has what lawyers call a jus quaesitum, or a right of ancient possession, which ought to stand good till it be overturned. If it cannot be proved that we always act from necessity, there is no need of arguments on the other side to convince us that we are free agents.” (Reid 1788, pt. IV, vi).

This weak or modest principle of common sense involves two essential elements: first, a certain conception of what counts as common sense beliefs and, second, an argument as to why this particular class of beliefs should enjoy a particular kind of default justification.

First, this metaphilosophy involves a certain conception of what counts as common sense beliefs. This conception is captured in the following definition – which is a stipulative definition (not a thesis about what has traditionally been meant by the phrase “common sense”):

Definition of common sense:

common sense = df the conceptual scheme and set of beliefs and principles that constitute our cognitive starting point.

Second, Thomas Reid’s quote also gestures at an argument for the view that this class of beliefs (the beliefs that constitute our starting point) enjoy a certain kind of privilege, a default justification. The idea here is that common sense beliefs enjoy a default justification in virtue of the fact that they constitute our starting point; this is so, as Reid puts it, because they are already in place at the start of the philosophical enquiry and therefore do not “need of arguments to convince us”, whereas the other side isn’t already in place and therefore needs arguments to convince us to change our view. The epistemic principle which is implicitly at work in this reasoning is a kind of “principle of epistemic inertia”: you need a special reason to change your beliefs (add new ones, or revise old ones), but you need no special reason to remain “epistemically inert” (to keep the same set of beliefs). This is a principle in “dynamic epistemology”, the epistemology of belief revision, or “change in view”. While a lot of work in epistemology concerns “synchronic” epistemology (the epistemology of belief states), there are some contemporary authors, in particular Gilbert Harman and Isaac Levi, who have been particularly attentive to the epistemology of belief revisions. Isaac Levi, for instance, formulates the same principle as follows:

“X should not modify his body of knowledge unless in doing so he improves it. Hence, even though X need not justify having h in his corpus of knowledge once he has accepted it, prior to doing so he will be under some obligation to justify adding h to his body of knowledge.” (Levi 1980, 1–2, emphasis mine).

When you combine this principle with the idea that common sense is the starting point, you arrive directly at the view that common sense enjoys a default justification (whereas beliefs not present in common sense have a burden of proof on their shoulders). This is clearly expressed in Gilbert Harman’s work (though not with the terminology of “common sense”):

“Your initial beliefs, plans, goals, and methods have an immediate default or prima facie “justification”. They are the “foundations” for your reasoning – foundations in the sense of starting points.” (Harman 2010, 153).Footnote 6

To summarize, the metaphilosophy of common sense I am relying on here is a very modest claim: it does not say that common sense beliefs are unassailable by philosophical arguments; it only relies on the idea that if some philosopher suggests that I should change my views, then she should provide reasons for that – and since common sense beliefs are (by stipulation) my starting point, a philosopher who wants me to change anything to my common sense framework has the burden of proof of providing (convincing) reasons to that effect (while remaining in the common sense view needs no particular argument).

The modest metaphilosophy of common sense that I am presenting here is very close David Lewis’ conception:

“I do respect common sense, within limits. […] It is far beyond our power to weave a brand new fabric of adequate theory ex nihilo, so we must perforce conserve the one we’ve got. A worthwhile theory must be credible, and a credible theory must be conservative. It cannot gain, and it cannot deserve, credence if it disagrees with too much of what we thought before. And much of what we thought before was just common sense. Common sense is a settled body of theory – unsystematic folk theory – which at any rate we do believe; and I presume that we are reasonable to believe it. (Most of it.) […] theoretical conservatism is the only sensible policy for theorists of limited powers, who are duly modest about what they could accomplish after a fresh start. Part of this conservatism is reluctance to accept theories that fly in the face of common sense.” (Lewis 1986, 134).

I will not spend more time here defending this metaphilosophy. In most occasions in which I have defended it, most philosophers accepted that it was a modest and very plausible principle of metaphilosophy – indeed so modest that the main objection I encountered was that it was not strong enough to count as a real “common sense philosophy”. I consider this objection as more historical than philosophical (“would historical common sense philosophers recognize their tradition of thinking?”), or merely verbal (“I don’t want to use the words ‘common sense’ to describe this metaphilosophy”). My intention here is to uphold the principle as I have stated it here, not to situate myself within a “historical tradition” nor to reclaim a particular word (“common sense”).

2 Common Sense vs Principles of Economy in Metaphysics

In the introduction, I opposed two methodologies in metaphysics: a methodology motivated by the preservation of common sense, and a methodology motivated by a principle of economy (or ontological parsimony). After having presented my preferred version of the metaphilosophy of common sense, I will now present the methodology of economy and discuss its compatibility or incompatibility with this metaphilosophy of common sense.

A standard formulation of the principle of economy in ontology (or Ockham’s razor) is as follows:

(OR) One should not accept entities in one’s ontology “beyond necessity” (i.e. without a positive and convincing reason to do so). (see Baker 2016, sec. 2)

In order to facilitate the comparison between the two methodologies, we could reformulate the moderate principle of common sense in a parallel fashion:

(CS) One should not make changes to common sense “beyond necessity” (i.e. without a positive and convincing reason to do so).

One difference between the two principles is that the first one concerns only ontological claims – claims about which (kinds of) entities exist, whereas the second principle concerns all sorts of claims. (Presumably not all beliefs in common sense are ontological. For instance, common sense arguably contains the belief that time has a direction, and that claim is not reducible to an ontological claim, although it may have ontological implications.) Therefore, the two principles do not cover exactly the same terrain, but inasmuch as common sense also contains ontological beliefs, the two principles will occasionally compete on the same ground.

That being said, it should also be obvious that there are some ontological claims for which the two principles will be in full agreement: these are the claims about entities that are not present in common sense. Consider for example scientific posits like electrons or ether: these entities are absent from the common sense picture of reality. Therefore, the metaphilosophy of common sense prescribes to accept their existence only if we have a positive and convincing reason to do so. Since there is such a positive reason for electrons, one should add them to the ontology of common sense. Since there is (now) no good reason to believe in ether, one should not add it to the ontology of common sense. The methodology of economy will deliver exactly the same result: this methodology requires a positive reason for accepting any entity in one’s ontology; therefore it requires a positive reason for adopting electrons and ether. What is true of scientific posits is also true of metaphysical posits. Consider for example the monotheistic idea of God. Arguably, this entity is not part of the ontology of the starting point of common sense. Therefore, (CS) will prescribe that if a metaphysician wants to have God in his ontology, he should have a positive and convincing reason to add this entity. And (OR) will make just the same prescription – not because God is absent of the common sense ontology, but because all entities require a positive and convincing reason to be accepted in one’s ontology.

On the other hand, there are cases in which the two methodologies will make different prescriptions. These are the cases of (types of) entities whose existence is granted from the start by common sense. Consider for example the existence of tables, chairs, trees and dogs. These are (types of) entities that are already present in common sense. Therefore, (CS) will tell us that we don’t need any particular positive argument in order to be justified in believing in their existence. On the contrary, anyone who denies their existence has to provide a positive and convincing argument because he wants us to move away from our epistemic starting point. Perhaps there are good philosophical arguments to abandon the existence of tables and chairs (see Van Inwagen 1990) or even the existence of trees and dogs (see Merricks 2006). And if a good argument of this kind is put forward, then (even according to the methodology of common sense) we will need to offer positive and convincing reasons to rebut these arguments. But in the absence of such arguments, the default position is to believe in the existence of tables, chairs, trees and dogs.

The principle (OR) makes a different prescription here: tables, chairs, trees and dogs, like any other entity, need a positive and convincing argument if we are to accept them in our ontology. The starting point or default position of such a methodology is an empty ontology, so that anything we accept in our ontology has to pay an entrance fee.

One way to present the opposition between (CS) and (OR) consists in distinguishing two kinds of economy principles, one of which is compatible with (CS), the other one being in conflict with (CS). I shall call them the Principle of common sense Economy (PCSE) and the Radical Economy Principle (REP).

(PCSE) One should not accept in one’s ontology entities that are not already accepted in common sense, unless one has a positive and convincing reason to do so.

(REP) One should not accept in one’s ontology entities for which one doesn’t have a positive and convincing reason to do so, even if they are already accepted in common sense.

In order to make the radicalism of (REP) more visible, we could also reformulate it in the following (equivalent) way:

(REP’) One should eliminate from one’s ontology all entities (even entities accepted by common sense) unless there is a positive and convincing reason not to eliminate them.

The methodology I will be relying on in the remainder of this paper accepts the modest form of economy (PCSE) but doesn’t accept the radical economy principle. More precisely, what I will try to show is that if we accept a methodology of common sense (or the economy principle consistent with this methodology), we should be led to endorse nominalism, while realism (contrary to common presentations) relies on a kind of radical economy.

Before I can establish this point, though, it is necessary to expose how I understand the dialectical situation of the debate between nominalism and realism.

3 The Dialectical Situation, and the Motivation of Realism

My main argument relies on a certain understanding of the dialectical situation of the debate about Universals. It is therefore necessary to make clear what this interpretation is. It involves the following four claims:

  1. (a)

    The thesis of realism is an existence claim (the existence of Universals).

  2. (b)

    This existence claim is offered as an explanation for some explananda.

  3. (c)

    The kind of explanation at play is metaphysical explanation or grounding.

  4. (d)

    The explananda of this explanation are predicative facts such as a is F.

Proposition (a) states that the debate between realists and universalists (as I understand it) is a debate about the existence of Universals, and not about the fundamentality (or groundedness) of Universals, as Schaffer (2009, 358, 362) and Schulte (2018) have recently defended. As we will see shortly, my adopting the traditional “existential” approach to the debate doesn’t mean that I reject altogether the importance of the notion of grounding in metaphysics. On the contrary, I believe that the notion of grounding is fundamental to understand the kind of explanation that motivates the realist’s claim – but that claim is an existential claim, not a grounding claim about undisputedly existing beings.

Proposition (b) states that the realist’s claim is defended as the proper response to a certain need for “explanation”. This proposition is so general that I don’t expect it to be controversial. The controversy of interpretation bears on what kind of explanation is looked for by the realist. Oliver (1996) has famously proposed three possible interpretations: the realist’s demand might be a demand (i) for a conceptual analysis of some sentences, (ii) for the ontological commitment of some sentences, or (iii) for the truthmakers of some propositions. At the end of the 1996 paper, Oliver concluded: “Moreover, the three interpretations of the problem of universals are the only senses I can give to the elusive idea of metaphysical explanation which is often invoked but never clarified. But perhaps I am missing something.” (Oliver 1996, 75). In the present situation of the debate in metametaphysics, it seems clear that the notion of metaphysical grounding should at least constitute a fourth candidate for the kind of “metaphysical explanation” the realist is looking for.

Proposition (c) is a take on what kind of metaphysical explanation is looked for. I follow Rodriguez-Pereyra (2000, sec. 3) in his rejection of conceptual analysis and ontological commitments as proper interpretations of what the realist is after. In 2000, Rodriguez-Pereyra considered that the only remaining candidate was the demand for truthmakers – which he reformulated as “truthmakers or ontological grounds”. For reasons I will develop in the next sections, I consider the explanation in terms of truthmakers as a subkind of the metaphysical explanation in terms of grounding (i.e. grounding in an existential fact).Footnote 7 Therefore, I will not reject altogether the possibility that the realist may be in search of a metaphysical grounding of the “truthmaker” variety, but what seems clear at least is that the realist is in search of grounding. That much seems clear, at least, for a contemporary reader accustomed to the grounding literature and the reflections on “in-virtue-of” claims in metaphysical explanations: the literature on universals before the emergence of the grounding debates is full of “in virtue of” assertions or questions (22 occurrences in Armstrong 1997). And I think Rodriguez-Pereyra is exactly right when he summarizes that responses to the problem of Universals are theories that “explain in virtue of what a single particular can have many properties” (Rodriguez-Pereyra 2000, 270 emphasis mine). Similarly, Michael Devitt has Armstrong asking the following question: “In virtue of what are a and b both F?” (Devitt 1980, 435 emphasis mine). I will assume that this search for “something in virtue of which…” is a search for metaphysical grounding. This interpretation of truthmaking in terms of grounding is disputable and has been disputed recently by (Audi 2019) but, as Paul Audi himself points out, this is still usually considered as a relatively standard interpretation.Footnote 8

Proposition (d) identifies the explananda of the metaphysical explanation. Alex Oliver (1996, 49) has shown that Armstrong was vacillating between the following explananda:

  1. (1)

    a and b are of the same type/have a common property F.

  2. (2)

    a and b are both F.

  3. (3)

    a and b have a common property, F.

  4. (4)

    a has a property.

  5. (5)

    a is F.

  6. (6)

    a has the property F.

Here, I follow Devitt (1980), Rodriguez-Pereyra (2000, sec. 5) and Imaguire (2018, sec. 5.2), for whom the real explananda are propositions of type (5), because the other propositions are either equivalent to (5) – this is the arguably the case for (4) and (6) – or explainable by propositions of type (5) – as Devitt (ibidem) argues, proposition (2) is true in virtue of a is F and b is F; and arguably (3) and (1) are equivalent to (2). Imaguire (ibidem) develops in more detail and in a more systematic way the same reduction, and does it explicitly in terms of grounding.Footnote 9 Therefore, in my understanding the question the realist is trying to answer when she appeals to the existence of Universals is the question “in virtue of what is the particular a F?”, and this is a question of grounding (which might or might not require a truthmaker, as we will discuss later).Footnote 10

To summarize, realists defend the existence of Universals, and they do so because they consider their existence as the required metaphysical grounding for predicative facts (expressed by propositions of the form a is F). The debate on Universals is the question whether they are right to posit such explanatory entities. But the more restricted problem I will try to illuminate is what kind of metaphysical methodology they are following when positing these entities, and in particular whether they are following a methodology of common sense.

4 The Burden of Proof According to (CS)

Now that we have a summary description of what realism and nominalism amount to, we can try and determine what theory would be favored by the common sense methodology presented in Sections 1 and 2.

The core principle of this methodology consists in saying the following: if a certain (type of) entity is present in the ontology of common sense, then the burden of proof is on the shoulders of those who deny the existence of this entity; conversely, if a certain entity is absent from the ontology of common sense, then the burden of proof is on the shoulders of those who affirm the existence of this entity.

The first thing we have to determine, then, is whether Universals (whatever theoretical arguments might exist for or against their existence) are present or absent in the ontology of common sense.

Some realists might be tempted to claim the advantage of common sense here. After all, is it not part of common sense that things share properties? If that much is part of common sense, then it would seem that (according to the common sense methodology), the burden of proof is on the shoulders of nominalists.

But the idea that Universals are part of the fabric of common sense ontology is far from obvious, and we can notice that Armstrong, even when he appeals to common sense doesn’t claim that common sense is already committed to Universals. What Armstrong considers as a datum of common sense, or as a “Moorean fact”, is “the fact of sameness of type” (Armstrong 1980, 441). In other words, the data of common sense are sentences of the form.

  • (1) a and b are of the same type/have a common property F.

  • (2) a and b are both F.

  • (3) a and b have a common property, F.

And Armstrong thinks that the account (or metaphysical explanation) of these commonsensical data requires one to posit the existence of Universals. He does not say that these Moorean facts reveal that the existence of Universals is already posited by common sense. That would indeed be a very bold claim since, as we have seen, these facts can be reduced to the more fundamental:

  • (4) a has a property.

  • (6) a has the property F.

which in turn can be reduced to

  • (5) a is F.

And it would be extremely implausible to claim that a commonsensical sentence of the form (5) manifests the presence of Universals in the ontology of common sense.

Maybe our “common sense realist” would have a better argument if she tried to show the presence of Universals in common sense starting with sentences containing abstract nouns. To take a famous list of examples (Armstrong 1978, 58; Lewis 1983, 348; Rodriguez-Pereyra 2002, 91):

  • (7) Red resembles orange more than it resembles blue.

  • (8) Red is a colour.

  • (9) Humility is a virtue.

  • (10) Redness is a sign of ripeness.

From the use of abstract nouns like “redness”, “colour”, “humility”, one might try to argue that common sense does contain in its ontology entities that are Universals.

But this would also be much too simplistic an argument: obviously, common sense doesn’t contain “stakes” in its ontology; arguably, it doesn’t contain “shadows” either. In order to show that something is an entity in the ontology of common sense, one cannot simply put forward substantival expressions of surface grammar.

The question whether sentences (7–10) can or cannot be paraphrased in a systematic way (see Lewis ibidem and Rodriguez-Pereyra ibidem) is not relevant here: this question is disputed by philosophers who wonder whether the truth of these sentences commits one to the actual existence of Universals. Our question here is completely different: we are wondering whether the commonsensicality of these sentences is a proof of the presence in common sense ontology of universal entities. To answer the latter question, it takes much more than the linguistic evidence of substantival expressions, whether or not we can paraphrase away these expressions. Here again, it is important to notice that Armstrong (or Pap, or Jackson), who put forward these sentences, never had the intention to conclude that Universals are already accepted in the ontology of common sense. Their intention was to show that sentences that are true (whether or not common sense accepts them) should lead us to posit the existence of Universals (when we reflect on what is required for them to be true).

As a matter of fact, I am not aware of realists who would have claimed that Universals are already present in the ontology of common sense. It seems to be a matter of consensus that Universals (if we accept them at all) are accepted as philosophical posits – something we add to common sense (for whatever philosophical reasons), but that is not present in common sense. Therefore, I will rely on this consensus as it is, and will not try to develop an argument for it. What I am really interested in, in this paper, is to determine what kind of methodology the realist is employing when she tries to add Universals to common sense.

To sum up: Universals are not present in the ontology of common sense; therefore, if we follow the common sense methodology, the burden of proof is on the shoulders of realists, who have to offer us positive and convincing arguments for the existence of these philosophical posits. But, of course, realists do offer positive arguments in favor of the existence of Universals – the central argument being the argument of the metaphysical grounding of predicative sentences, as we saw in Section 3. Therefore, for the time being, we cannot say that the methodology of common sense precludes the possibility of becoming a realist. It does give an initial advantage to the nominalist, but the realist might eventually win the game. What remains to be done, therefore, is to determine what kind of methodology is used by the realist when she offers her argument about the grounding of predicative sentences. We will not be discussing the question whether this argument is convincing or not; what we want to know is not whether realism wins the game; we want to know whether her strategy follows a methodology of common sense or not. And in the next section, I will argue that the argument about the grounding of predicative sentences does not follow a methodology of common sense, but rather a methodology inspired by a motivation of radical economy.

5 The Realist’s Motivation as Involving a Radical Form of Ontological Economy

Whatever methodology is adopted by the realist, she will have to provide a positive and convincing reason for accepting the existence of Universals – if she adopts the methodology of common sense, this is because the Universals are not present in the ontology of common sense and therefore need motivation; but if she adopts instead the methodology of radical economy, she needs to motivate the acceptance of Universals for an even more trivial reason: in that methodology, any entity needs a positive motivation before we add it to our ontology.

Let us come back then to the main motivation of realists. Armstrong calls it “the One over Many argument” (Armstrong 1978), because the main question to which the realist is seeking an answer is: “in virtue of what are a and b both F?” (Devitt 1980, 435). If we read this “in virtue of” as a question of metaphysical explanation or metaphysical grounding, the issue then is to find a fact that might ground the fact that “a and b are both F”. As Devitt (ibidem) argued, the first step of the response is rather obvious: a and b are both F in virtue of the facts that “a is F”, and that “b is F”. This much should not be problematic. If there is a problem then, a problem that would motivate the postulation of Universals, it must be the problem of grounding predicative facts, i.e. facts of the form “a is F” or “b is F”. That’s why I follow (Devitt 1980), (Rodriguez-Pereyra 2000) and (Imaguire 2018) in saying that the core motivation of realists is the explanation of predicative facts (and not, fundamentally, the problem of grounding “resemblance facts” per se). As Devitt clearly summarizes:

“‘In virtue of what is a (or b) F’. If the One over Many argument poses a problem it is this. That was historically the case and, though Armstrong always states the problem in terms of identities in nature, it is the case for him too.” (ibidem).

We can easily reformulate Armstrong’s description of the debate in terms of how the various doctrines respond to this question: “In virtue of what is a F?”, or equivalently “what grounds facts of the form ‘a is F’?” If we set aside trope theory, the three answers to this question are: Class Nominalism, Resemblance Nominalism, and Realism.

Let us start with the grounding relations as they are understood by the realist. There are two levels of grounding for the realist. The first level grounds resemblance facts in predicative facts:

(11) “a resembles b [or is of the same type as b]” in virtue of “a is F” and “b is F”

The second level, in turn, concerns the grounding of predicative facts themselves, or that “in virtue of which a predicate applies” to a particular. It is at this second stage that universals appear. Armstrong often expresses himself saying that a predicate applies to some particular “in virtue of some universal F” – which I interpret as meaning “in virtue of the existence of some universal F”.Footnote 11 If we add that a predicative fact like “a is F” is also partly grounded in the existence of a itself, we arrive at the following grounding claim:

(12) “a (or b) is F” in virtue of “there exists a particular a (or b)” and “there exists a universal F-ness”

The most important element here is of course the second half of proposition (12), i.e. that “a is F” is grounded not only in the existence of a particular but also in the existence of a Universal, and that’s how the problem of grounding predicative facts commits us to the postulation of Universals.Footnote 12

The Resemblance nominalist also offers a grounding for predicative facts, but in order to do so, he reverses the order of grounding of proposition (11). His central claim is that:

(13) “a is F” (and “b is F”) in virtue of the fact that “a resembles b”

In other words, predicative facts are grounded in resemblance facts, and resemblance facts are taken to be brute (or fundamental) facts.

Finally, the Class nominalist proposes to ground predicative facts in facts about class membership:

(14) “a is F” (and “b is F”) in virtue of the fact that “a belongs to class C” (and so does b)

All three theories offer their own account about how to ground predicative facts. Predicative facts are grounded (respectively) in facts about the existence of Universals (12), resemblance facts (13) or facts about class membership (14).

There is another version of nominalism, inspired by (Quine 1953), which Armstrong has labelled “Ostrich nominalism”. According to the Ostrich nominalist, there is nothing in virtue of which a is F, no facts that ground predicative facts: “we have nothing to say about what makes a F, it just is F; that is a basic and inexplicable fact about the universe.” (Devitt 1980, 436). For Armstrong, this reaction is comparable to the Ostrich who digs her head in the sand in order not to see an obvious problem. Or to take another Armstrongian metaphor: the Ostrich nominalist is one who refuses to answer “a compulsory question in the examination paper” (Armstrong 1978, 17). The compulsory question is of course: “what grounds predicative facts?” Strictly speaking, the Ostrich nominalist does give an answer: the answer is: “nothing” (predicative facts are ungrounded or fundamental facts). But that doesn’t count as an answer for Armstrong who sees it as a compulsory task of the metaphysician to provide a positive ground for predicative facts.

The Ostrich nominalist will accept the first half of the realist’s grounding structure, i.e. he will accept that:

(11) “a resembles b [or is of the same type as b]” in virtue of “a is F” and “b is F”

But he will stop the grounding regress there: for him, “a is F” and “b is F” are not grounded.

If we compare the various theses in the debate according to their account of the grounding of predicative facts, we can easily understand why Armstrong sees Ostrich nominalism as a non-starter: we have three candidates who offer a positive grounding for predicative facts, and a fourth candidate (the Ostrich) who just refuses to offer an explanation where all others can give one. If things are presented in this way, the Ostrich nominalist seems to have a fatal flaw.

But this presentation of the debate is unsatisfactory: why should we concentrate all the debate on the grounding of predicative facts? The debate between these philosophers is a debate between various grounding structures – some with several storeys, some with just one storey. What we want to know is which of these grounding structures is the best, not just with respect to how it grounds predicative facts, but in general and as a whole. Now, when we compare these grounding structures as wholes, we immediately notice that there is something they all have in common: all of them stop the grounding regress somewhere, i.e. they all accept brute or fundamental facts. Granted, the Ostrich nominalist considers predicative facts as brute or fundamental facts. But as Devitt rightly asked: “Why be dissatisfied with this? Explanation must stop somewhere.” (Devitt 1980, 436). The only difference with the other solutions is that they stop elswhere: the Resemblance nominalist accepts Resemblance facts as fundamental; the Class nominalist accepts facts about class membership as fundamental; and the realist accepts existence facts (of Universals, and of particulars) as fundamental. Therefore, Ostrich nominalism cannot be criticized for accepting brute facts. The real question that should be asked is: which facts are better candidates to stop the grounding regress? And in particular: is there anything wrong in stopping the regress with predicative facts instead of stopping it elsewhere?

In the remainder of the paper, I will set aside Class nominalism and Resemblance nominalism and compare only Ostrich nominalism with realism. The reason is that Class nominalism and Resemblance nominalism, independently of the place where they choose to stop the grounding regress, seem to have an intuitive defect in their overall grounding structure. As we saw earlier, the Resemblance nominalist reverses the intuitive:

(11) “a resembles b” in virtue of “a is F” and “b is F”

into

(13) “a is F” and “b is F” in virtue of the fact that “a resembles b”

If we set aside any other consideration, it seems clear that the initial intuition goes rather in favour of the order of grounding (11) and not (13), as Van Cleve has convincingly argued (Van Cleve 1994, 579). Why then does the Resemblance nominalist choose to reverse the intuitive order? Presumably because she feels obliged to find out a fact that might ground predicative facts. And instead of positing new entities in the ontology, she prefers to ground predicative facts in some uncontroversial facts, namely Resemblance Facts, who have the advantage that they can be shown to stand in modal equivalence with predicative facts. In other words, the main motivation for reversing the intuitive order of grounding (11) is that they rule out from the start the possibility of accepting predicative facts as brute facts. But if Ostrich nominalism were not ruled out from the start – i.e. if we didn’t consider it obligatory to offer a positive grounding for predicative facts – then it would certainly be better not to reverse the intuitive order of grounding and to keep (11) instead of (13).

Class nominalism is counter-intuitive in a similar way. Instead of the intuitive claim that:

(15) “a and b belong to class C” in virtue of the fact that “a is F” and “b is F”

the Class nominalist reverses the order of explanation and says:

(14) “a is F” and “b is F” in virtue of the fact that “a and b belong to class C”

And the reason for this reversal seems to be that the Class nominalist considers it necessary to ground predicative facts in something. But if this were not taken as a necessary demand, then it would seem better to maintain the intuitive order of explanation.

Realism and Ostrich nominalism are the two theories that keep the intuitive order of explanation between “a resembles b” and “a (or b) is F”. Their difference concerns only the facts at which they stop the regress of explanation: Ostrich nominalism stops the regress at predicative facts, while realism goes one step further and stops at existence facts (of Universals, and of particulars). The question then is: which of these two ways of stopping the regress is better. For someone who adopts the methodology of common sense, we have seen that there was a presumption in favour of nominalism; therefore, in the absence of a positive and convincing reason to prefer any of these two ways, the nominalist will win the game by default. In this context, then, the question becomes: is there a positive and convincing reason to reject the Ostrich way of stopping the regress (at predicative facts) and posit Universals in order to stop the regress one step further?

I believe the realist does have some kind of argument or motivation to prefer her way of stopping the grounding regress, even though the dialectical situation is not typically presented in this way. My understanding of her motivation is that she would prefer to reduce predicative facts to existence facts, and that this preference comes from her formulation of the Truthmaker principle. Let me explain.

As Lewis (1992, 218) has shown, the demand for truthmakers contains two elements, one of which is uncontroversial, the other being quite disputable. The uncontroversial aspect of the truthmaker principle is that “truth is supervenient on Being” – to use Bigelow’s formulation (Bigelow 1988, 132), approved by Lewis (ibidem). A somewhat different formulation, but equivalent for our purposes is the following:

(TM1) Facts about what propositions (or sentences) are true are grounded in facts “in the world”, i.e. facts about things.Footnote 13

But, as Lewis says, facts about things are not limited to facts about whether things are (or what things exist); they include facts about how things are:

“As an Ostrich […] I want to construe ‘being’ broadly: it covers not only whether things are, but also how they are.” (Lewis, ibidem).

But defenders of the Truthmaker principle typically want to go further and say that:

(TM2) Facts about what propositions (or sentences) are true are grounded in facts about whether things are (i.e. in existence facts).

This is, according to Lewis, the controversial aspect of Truthmaking. And it is clearly present, for example, in Armstrong’s definition of the truthmaker principle:

“for every contingent truth, there must be something in the world that makes it true. […] The ‘making’ is not causality, of course: Rather, it is that in the world in virtue of which the truth is true.” (Armstrong 1989, 88).Footnote 14

The “in virtue of” phrase shows that Armstrong’s principle is a demand for grounding, but the grounding he is looking for seems to be a thing, rather than a fact or a proposition. One way to interpret this is to read the grounding relation here as a relation between (grounded) facts and (grounding) objects. But in fact, the most natural reading is that what does the grounding is the existence of the corresponding object. In other words, as Bliss and Trogdon have proposed, it is quite straightforward to propose the following reading:

“x is a truthmaker for the proposition that p just in case [x exists] grounds [p]” (Bliss and Trogdon 2016, sec. 6.3)

Now let us compare the uncontroversial (TM1) and the controversial (TM2) (in a slightly revised but equivalent formulation):

(TM1’) Facts about what propositions are true are grounded in facts about things (whether things are – existence facts – or how things are – predicative facts).

(TM2’) Facts about what propositions are true are grounded in facts about whether things are – existence facts.

(TM2’) entails (TM1’), but not vice versa – (TM2’) is strictly stronger than (TM1’).Footnote 15 Therefore, what renders (TM2’) controversial must be the element that it adds to (TM1’). But how do we get from (TM1’) to (TM2’)? What is added by (TM2’)? My understanding is that defenders of the Truthmaker principle are implicitly accepting the following principle:

(TM3) Facts about things in general (existence facts and predicative facts) are all ultimately grounded in existence facts.

Because Grounding is (commonly taken to be) transitive, if you accept both the uncontroversial (TM1) and the principle (TM3), you can derive the truthmaker principle (TM2). Principle (TM3) implies that all facts whatsoever must ultimately be grounded in existence facts. And if my interpretation is correct, such a principle is implicitly what motivates the realist when she refuses to stop the regress of explanation at predicative facts (as the Ostrich nominalist does) and insists in looking for a thing the existence of which would ground predicative facts (namely a Universal).

But why does the realist want to ground all facts in existence facts, instead of keeping (as Lewis proposes) existence facts and predicative facts on a par, as the ultimate ground for all truths? What is the advantage of reducing all facts to existence facts instead of existence facts plus predicative facts? The answer seems clear: this is an advantage of economy. Since everyone (even the Ostrich nominalist) must accept at least existence facts as fundamental facts, it is more elegant and more economical to accept just those and to eliminate the exuberant fauna of brute predicative facts. My interpretation can be summed up as follows:

(Economical realism) The central motivation of the realist, when she refuses to stop the regress of explanation at predicative facts, is a motivation of economy: to restrict herself to just one kind of brute facts, i.e. existential facts.

This economical motivation is comparable with the advantage that Rodriguez-Pereyra sees for Resemblance nominalism as opposed to Ostrich nominalism. Rodriguez-Pereyra considers the former as more economical than the latter, for the following reason:

“Resemblance nominalism resembles Ostrich nominalism. The difference is that whereas the latter admits many sorts of basic facts involving only particulars – ‘a is scarlet’, ‘b is an electron’ – the latter admits only basic facts of the form ‘a resembles b to such and such a degree’.” (Rodriguez-Pereyra 2019, sec. 4.1)

Saying that the core motivation of realism (the One over Many argument, or the argument about the grounding of predicative facts) is a motivation of economy is not sufficient to prove that it is in tension with the methodology of common sense. We have seen in Section 2 that one version of the economy principle (PCSE) is completely compatible with a common sense metaphilosophy.

Therefore, the final question we have to ask is whether the economy principle on which (Economical realism) is relying is an economy principle compatible with a common sense methodology, or rather a radical economy principle (REP). The answer is that (Economical realism) relies on a radical form of economy, a motivation to eliminate from the fundamental ground of reality as many entities as possible, even if these entities are already part of common sense. Predicative facts, like existence facts, are part of the fabric of common sense; therefore, a methodology of common sense will see no need to try and dispense with them or eliminate them from the fundamental ground of reality. The attempt to eliminate them (and keep only existence facts) comes from the “taste for desert landscapes”, not from the idea of doing as few modifications as possible to common sense. This is a motivation which could lead to eliminate common sense data for the purposes of radical economy. Therefore, a philosopher who would follow a common sense methodology in metaphysics should not accept the core argument in support of the existence of Universals.

To summarize, not only is it the case that a common sense methodology puts the burden of proof on the shoulders of the realist (because she posits entities that are not already present in the ontology of common sense), but more importantly, the argument that the realist employs to motivate her acceptance of Universals – namely the argument from the grounding of predicative facts – implicitly relies on a methodology of radical economy, incompatible with the methodology of common sense. Therefore, contrary to a common presentation of the debate, a common sense methodology should lead one to remain a nominalist (and more precisely to adopt Ostrich nominalism), while realism is motivated by a principle of ontological economy. (Of course, this is compatible with saying that other philosophers motivated by radical economy could interpret differently the demands of radical economy and also arrive at nominalism, for opposite reasons.)

6 Objections

I will consider two objections to the argument of Section 5, i.e. to my interpretation of realism as relying on a motivation of (radical) ontological economy.

Objection 1: “You cannot say that the grounding structure of realism grounds everything on existence facts. The fact that ‘a is F’ cannot be completely grounded in a’s existence and Fness’s existence, because a and F could exist without a’s instantiating Fness. Therefore ‘a is F’ must also be partly grounded in some tie between a and F, which according to Armstrong (1989, chaps. 5, IV) is the state of affairs ‘a’s being F’ or ‘a’s instantiating Fness’. But if that is true, then it seems that the brute facts that are accepted by the realist include not only existence facts but also instantiation facts that are not purely existence facts. Therefore, the realist cannot be motivated by the purpose of reducing all facts to existence facts, which in turn means that she may not be motivated by ontological economy.”

Response: One way to respond to this objection is to emphasize Armstrong’s attempt to ground predicative facts in the existence of states of affairs: when Armstrong posits states of affairs in order to solve the problem of the tie, it is precisely in application of the truthmaker principle; the state of affairs is posited as a truthmaker, i.e. as a thing the existence of which grounds predicative facts. Therefore, according to the most natural reading of Armstrong’s strategy here, the realist’s grounding structure is one which is ultimately grounded in only existence facts: existence of particulars, existence of universals and existence of states of affairs.Footnote 16 For simplicity, the previous section considered only existence facts about particulars and universals, but adding existence facts about states of affairs does not change the general conclusion, namely that the realist’s fundamental motivation is to ground all facts in existence facts.

But suppose we interpret Armstrong differently. Suppose we consider that for him the non-relational tie of instantiation introduces a non-existential fact at the basis of the grounding structure. In other words, “a is F” would be grounded in two existential facts “a exists” and “Fness exists”, plus the non-existential fact that “a instantiates Fness”. I don’t think this interpretation is the proper one, but even in this case, realism would have the same kind of advantage over Ostrich nominalism that Rodriguez-Pereyra saw in Resemblance Norminalism (cf. Section 5), namely that realism would ground all facts in only one kind of non-existential fact (x instantiates Pness) whereas Ostrich nominalism has as many kinds of basic facts as there are properties in realism. Therefore, even in this interpretation, it would remain plausible to say that realism relies on a motivation of economy – an economy of kinds of basic facts.

Objection 2: “Realism does not eliminate predicative facts; it grounds them, but they still exist, as grounded entities. Therefore, realism cannot be presented as eliminating entities present in common sense. And, arguably, in considering predicative facts as derivative, it makes an addition to common sense, but no revision of common sense, unless you want to claim that predicative facts are fundamental according to common sense, and that realism revises this status by making them grounded and derivative. But do you really want to make this claim?”

In general, I think common sense can contain intuitions about what is more or less fundamental. I am inclined to think, for example, that it is a matter of common sense that “the singleton {Socrates} exists in virtue of the existence of Socrates” and not vice versa. And I am also inclined to think that common sense does not contain a general pattern for the grounding of predicative facts. But that is quite different from saying that common sense contains the view that predicative facts are not grounded, or fundamental, and I am not inclined to make such a claim. Now, if I admit that common sense is compatible with predicative facts being grounded, then it is right to say that realism, in making them grounded, does not impose a revision of common sense. But that is not the important point here. The important point is the motivation the realist has for grounding predicative facts. Since common sense contains both existential facts and predicative facts, why should anyone want to make only the former fundamental (and the latter grounded in them) rather than having both as fundamental facts? My response is that the motivation here is a kind of economy principle which is incompatible with the methodology of common sense. Here is why. The attempts by many philosophers to have as few entities as possible at the fundamental level are clearly motivated by considerations of economy, because of the idea that derivative entities have less “theoretical cost” than fundamental ones, or even no cost at all. This principle is formulated very explicitly by Jonathan Schaffer, who calls it “the Laser”:

“By the lights of The Laser, derivative entities are an ‘ontological free lunch’, in the sense that they are genuinely new and distinct entities but they cost nothing by the measure of economy. The Laser thus incorporates an implicit distinction between the commitments of a theory, and the cost of such commitments. (…) Derivative entities cost nothing further, beyond the cost incurred for positing their fundamental grounds.” (Schaffer 2015, 647–48)

Accordingly, the attempt to make predicative facts derivative instead of basic – and to have only existential facts as basic – is motivated by the intention to “pay the cost” only for the latter but not for the former. Is this a motivation consistent with the methodology of common sense? The answer is “no”. According to the methodology of common sense, entities already present in common sense don’t need anything in order to be added to our theory; they are already there, without any need for a special motivation. In terms of ontological costs, this means that entities already present in common sense come with no cost at all; they are already an ontological free lunch. For that reason, if you accept a methodology of common sense, it makes no sense to try and “diminish” the theoretical cost of common sense entities (they don’t have any cost), or to try and make them an ontological free lunch (they already are). This attempt can make sense only if you adopt a methodology in which all entities (including common sense entities) need to be positively motivated before we accept them in our ontology. This is the definition of the methodology of radical economy.

Therefore, even if realism doesn’t eliminate predicative facts altogether, and only eliminates them from the fundamental level of reality, the main conclusion remains untouched, i.e. that this elimination is motivated by a methodology of radical economy, incompatible with a methodology of common sense.

7 Conclusion

A common presentation of the debate about Universals considers that a common sense methodology backs the realist’s position, while the methodology of economy favours nominalism. I have tried to show that this presentation is wrong and that a philosopher adopting a methodology of common sense should rather be an (Ostrich) nominalist, while the central motivation for realism is in fact a motivation of theoretical economy.

My argument has two essential steps.

The first step is to show that, according to a methodology of common sense, entities that are not present in common sense need a positive and convincing reason before we adopt them into our ontology. Since Universals are absent from the ontology of common sense, this methodology puts to the burden of proof on the shoulders of the realist. By default, in the absence of a convincing argument for Universals, the common sense philosopher will remain a nominalist.

The second step analyses the argument provided by the realist in defence of the existence of Universals. Following Rodriguez-Pereyra, I have interpreted Armstrong’s central argument (the One over Many) as an attempt to find truthmakers for predicative truths; and this, in turn, is motivated by an attempt to ground all facts in just one kind of facts (existential facts). My interpretation is that this (strong) truthmaker principle is motivated by a desire to reduce the theoretical cost of the entities we accept in our ontology. In other words, the central argument for Universals is in fact motivated by a principle of economy. And even though some economy principles are compatible with the methodology of common sense, the economy the realist is trying to make here is a radical form of economy, incompatible with the methodology of common sense, because it is an attempt to “reduce the cost” of entities that the methodology of common sense already considers as being an ontological free lunch.

Therefore, a philosopher who follows an (even modest) methodology of common sense should not adopt realism about universals. He should remain a nominalist. More precisely, he has no reason to try and ground predicative facts, and should consider them as basic since we have no motivated theory of how to ground them. In other words, he has all reasons to be an Ostrich nominalist.

Conversely, the philosopher who arrives at realism is motivated by a methodology of (radical) economy. This is not to say that a philosopher motivated by economy will necessarily be a realist. I have not tried to establish this conclusion, and for all I know, there might be other, more convincing ways to apply the radical economy principle, and arrive at nominalism instead of realism. But if you are to defend realism, the common sense methodology is not the proper way to go. Common sense philosophers should be Ostrich nominalists.