Metaethics and the Nature of Properties

Abstract

This paper explores connections between theories of morality and theories of properties. It argues that (1) moral realism is in tension with predicate, class and mereological nominalism; (2) moral non-naturalism is incompatible with standard versions of resemblance nominalism, immanent realism and trope theory; and (3) the standard semantic arguments for property realism do not support moral realism. I also raise doubts about trope-theoretic explanations of moral supervenience and argue against one version of the principle that we should accept theories that maintain neutrality.

I

Wilfrid Sellars claimed that ‘The aim of philosophy, abstractly formulated, is to understand how things in the broadest possible sense of the term hang together in the broadest possible sense of the term’ (1963, p. 1). My aim here is to consider how two particular things hang together: our understanding of morality and our understanding of properties.

To begin, distinguish two questions. First: what are the logical and pseudo-logical connections between theories of morality (M₁, M₂, …) and theories of properties (P₁, P₂, …)? Here logical connections include entailment and consistency. By ‘pseudo-logical connections’ I mean those weaker relations—such as coherence, evidential support and explanatoriness—that are often considered relevant to assessing the plausibility of collections of claims. This first question most obviously concerns connections between theories but extends easily to arguments. So, for example, we can ask whether the standard arguments in favour of P₁ support M₁. I consider this question in §iv.

Second, supposing that we have established some logical or pseudo-logical connections between theories and arguments in one area and another, how should we assess the bundles thus connected? For example, suppose there are reasonable arguments in favour of P₁ and M₁, but we have established that P₁ entails the falsity of M₁. How should we proceed? Modus ponens or modus tollens? A third option is to remain agnostic, accepting only that the combination has been ruled out. Another example: suppose there are plausible arguments in favour of two incompatible theories in some area, M₁ and M₂. And suppose that, in a distinct yet related area, there is only one theory—P₁—that is consistent with both M₁ and M₂. Should we accept P₁? Jussi Suikkanen’s (2024) paper offers a positive answer to this question, at least where P₁ is the trope theory of properties and M₁ and M₂ are metaethical naturalism and non-naturalism, respectively. I offer an alternative view in §v.

What follows addresses these two questions in order. My approach is piecemeal, highlighting a few threads that others might later hang together more comprehensively.

II

It will help to know the content of some theories of morality and properties. Let us begin with the latter.

Here is my mug, next to my glasses. Both are blue, solid, and part of my office. What is going on? The aim of ontology is to give an account of the fundamental constituents of a real situation like this, ideally one that will be generalizable to all parts of the universe (Campbell 1990, p. 1).

It is natural to distinguish concrete objects, such as my mug or my glasses, from their features, such as blueness, solidity and parthood.¹ The former are particulars, the latter properties.² The first part of the job description for properties, therefore, is to furnish the natures of things.³ A second role concerns similarities: properties seem to be ways in which things can be genuinely resembling. For instance, my mug and my glasses share a genuine similarity: they are both blue. A third role concerns predication. Some predicates—such as ‘blue’—seem to attribute genuine features to objects. Properties are good candidates to be the things denoted by such predicates. On this view, just as singular terms like ‘Neil’s mug’ denote particular objects, predicates like ‘blue’ denote properties, with the copula ‘is’ enabling us to say that the latter characterizes the former.

These roles for properties are presented in order of increasing controversy. Not all parties accept that there are fundamental constituents of reality, or any thing at all, that plays these roles. Nevertheless, the phenomena here—natures, similarities, predication—are undeniable. An account of properties is an account of the nature of what, if anything, is the best candidate to play these roles and of the phenomena that underlie them.

On to the theories. The first—ostrich nominalism—denies that properties exist. On this view objects, like my mug, have natures such as being blue, however this state of affairs does not require the existence of an entity—blueness—that somehow inheres in the mug. Rather, the fact that my mug is blue can be taken as a basic fact about the world, requiring only the existence of the mug. Ostrich nominalists thus understand predication as follows:

(g) The sentence ‘a is F’ is true if and only if there exists an x such that ‘a’ designates x and ‘F’ applies to x.

The fact that ‘F’ applies to x is explained by the fact that x is F, which requires no existent corresponding to ‘F’ (Devitt 1997, p. 96). Extending the account, ostrich nominalists hold that all that is required to make claims like ‘a and b are both F’ true is the truth of the claims that ‘a is F’ and ‘b is F’. And, given (g), neither of these commits us to the existence of properties. One underlying principle here is the Quinean standard of ontological commitment: we are committed to the existence of all and only those entities required to make the sentences we affirm true (Quine 1997, pp. 82–5). Since properties are not required for the truth of claims like ‘Neil’s mug is blue’ and ‘Neil’s mug and glasses are both blue’, we are not ontologically committed to them.

A second view—predicate (or concept) nominalism—accepts that properties exist but holds that they are merely the ‘semantic shadows’ of predicates (or concepts). So, for an object to possess a property is for it to fall under the corresponding predicate. For two objects to be similar is for them to fall under the same predicate. The relation of ‘falling under’ is taken as primitive (Armstrong 1978, vol. I, pp. 11–27; Schiffer 2003).

A third view—class nominalism—takes properties to be sets or classes of particulars (Lewis 1997a).⁴ So, for an object to possess a property, for example, blueness, is for it to be a member of the class of blue things. For two objects to be similar in respect of some property is for them to be members of the same class. On this view, a sentence of the form ‘a is F’ is true if and only if a is a member of the class of things that are F. Possessing a property is explained in terms of class membership, not vice versa. Class membership is primitive, that is, requires no further existents in order to obtain. Class nominalists typically take classes to be mind- and language-independent, abstract (that is, non-spatio-temporal), and generated without restriction: any collection of particular things forms a class. Properties are therefore abundant: there are as many as there are classes of things, and many will not mark genuine similarities between their possessors. There is, for example, a property corresponding to the class of objects whose sole members are David Bowie, Averroes’ last meal, and the number seven. If we stipulate that ‘a is BAS’ is true if and only if a is a member of this class, then this property—BASness—is shared exclusively by Bowie, Averroes’ last meal, and the number seven, despite a lack of genuine similarity between them. Class nominalists typically take properties to be classes of actual and possible particulars, where the latter are understood as genuine existents, inhabiting non-actual yet concrete possible worlds (that is, class nominalists are typically modal realists). This is to allow for the possibility of properties without any actual instances, and to explain the distinctness of properties that, in the actual world, have exactly the same instances (Lewis 1997a, p. 173).

A fourth view—mereological nominalism—claims that properties are mereological wholes or fusions of particulars (Effingham 2018). Whereas a class of concrete particulars is abstract, and particulars are members, a mereological whole of concrete particulars is concrete, and particulars are parts. Mereological wholes are therefore located where all of their parts are located. On this view, for an object to possess a property is for it to be part of a mereological whole. For two objects to possess the same property is for them to be parts of the same whole. A sentence of the form ‘a is F’ is true if and only if a is a part of the mereological whole of F things. Possession of properties is explained in terms of parthood of wholes, not vice versa, and parthood is primitive. Mereological nominalists also typically accept unrestricted mereological composition: any two objects compose a further whole. Thus, as with class nominalism, they take properties to be abundant: there are as many as there are mereological wholes, and most will mark no genuine similarity between particulars. Again, as with class nominalism, mereological nominalists typically take properties to be constructions from actual and possible particulars, and for the same reasons.

A fifth view is resemblance nominalism. On this view a particular object, such as my mug, possesses the property of blueness just in case and because it resembles all other blue things (Rodriguez-Pereyra 2001). Objects share a property in virtue of resembling each other, and resemblance facts are primitive. A sentence of the form ‘a is F’ is true if and only if a resembles all other F things. Resemblance nominalism can (but need not) be understood as a form of class nominalism, according to which properties are classes of resembling particulars. Unlike other versions of class nominalism, this restriction allows resemblance class nominalists to hold that only some classes of particulars correspond to properties, maintaining the link between sharing a property and genuine similarity.

All nominalists deny the existence of universals. Property realists, on the other hand, identify properties with universals. A universal is an entity that can have multiple instances, remaining identical between them (Daly 1997, p. 140).⁵ Universals also mark genuine similarities between their instances. The universal blue, for instance, has one instance in my mug and one in my glasses, and this explains the similarity between them. Objects possess properties in virtue of instantiating the related universal. A sentence of the form ‘a is F’ is true if and only if there is a particular, x, such that ‘a’ designates x, a universal ɸ such that ‘F’ designates ɸ, and x instantiates ɸ. Varieties of realism are distinguished by their account of the nature of universals and instantiation. On one view—transcendental realism—universals are abstract, mind-independent entities that particulars partake in or imperfectly resemble (Russell 1912, ch. 9). Transcendental universals themselves are causally inert (because abstract) but their instances can be causally efficacious. On another view—immanent realism—universals are wholly located at each of their instances, which can be causally efficacious (Armstrong 1978, vol. II). On this view, to say that a particular instantiates a universal is to say that they are bound together in a state of affairs (such as my mug being blue). Such states are the basic furniture of the world, and particulars and universals are their co-dependent constituents (thus there are no bare particulars or uninstantiated properties). On both versions of realism, instantiation is taken as primitive (Edwards 2014, pp. 24–6).

Trope theory is an alternative to both nominalism and realism. According to it, the fundamental constituents of reality are tropes. A trope is a particular (that is, unrepeatable) entity of a specific nature, such as being blue. We can pick out tropes by using phrases such as ‘the blueness of my mug’, so long as we are not misled by this to suppose that the trope is a concatenation of two types of thing, universal and particular (Schaffer 2001, p. 249). Rather, tropes are simple (not made up of other types of thing) and fundamental (other categories are constructions of them). Tropes are also abstract, not in the sense of being non-spatio-temporal, but in the sense that ‘they (ordinarily) occur in conjunction with many other instances of qualities … and that, therefore, they can be brought before the mind only by a process of selection’ (Campbell 1990, p. 3). Tropes can also resemble each other (or fall into resembling classes) with this resemblance (or membership) taken as primitive (Williams 1997, p. 116; Campbell 1990, pp. 30–40; Ehring 2011).

According to trope theory, concrete particulars and properties are constructions of tropes. A concrete particular is a concurrent sum of tropes, for example, a blueness trope, a solidity trope, a mug-shape trope and so on, all bundled together. A property is a set of resembling tropes. Blueness, for instance, is the set of resembling blue tropes. For an object to possess a property, F-ness, is for a member of the set of resembling F tropes to be part of the concurrent bundle of tropes that constitute that object. For two objects to share a property is for a trope that is part of one concurrent sum of tropes to resemble a trope that is part of a second concurrent sum of tropes. And the sentence ‘a is F’ is true if and only if the concurrent sum of tropes, a, includes a trope which is a member of the resembling set of F tropes (Williams 1997, p. 119; Campbell 1990, p. 41).

These, then, are some of the options we have for an ontology of properties.⁶ In the next section, I turn to metaethics, to consider which of these views best fits moral realism.

III

Our moral practice involves the phenomena of thinking and speaking in distinctively moral terms, terms like ‘morally wrong’, ‘morally good’ and ‘morally virtuous’. The aim of metaethics is to give a systematic account of these kinds of phenomena, ideally one that will be generalizable without theoretical stress to all elements and types of moral practice.

The most popular metaethical theory is moral realism. Moral realists hold that moral judgements cognize mind-independent moral reality. On this view, moral thoughts are beliefs with moral representational contents. Indicative moral sentences represent reality in moral ways. Sincere utterances of such sentences serve to express such beliefs, thereby offering moral descriptions of the world. Moreover, some of these beliefs are true and made true by a mind-independent moral reality. The denizens of this reality are typically taken to include moral properties.⁷ The standard view, therefore, is that a moral predicate like ‘wrong’ denotes the moral property of wrongness, and to assert that ‘murder is wrong’ is to offer a representation of the world as one in which that property is possessed by the object denoted by the term ‘murder’. Moral predicates therefore have a property-ascribing function (Cuneo 2014, p. 241; Shafer-Landau 2003, p. 83 n. 36). Realists disagree amongst themselves about the nature of moral properties, for example, whether they are natural, supernatural or non-natural (which, for present purposes, I take to mean at least non-causal), and whether they are reducible to other types of property. They agree, though, that we need to postulate moral properties to explain the truth of moral predications, and to function as the referents of moral predicates.

Which theory of properties best coheres with the property-invoking claims of moral realism?

First, the properties moral realists invoke cannot be properties as conceived by predicate, class or mereological nominalists. The quickest way to see this is to note that properties as conceived by these theories need not mark genuine similarities between particulars, whereas moral realists hold that moral properties do mark such similarities. For example, according to class nominalism, BASness is a property, but there is no genuine similarity between all and only the three objects that are BAS. Realists hold, though, that there is a genuine similarity between all and only wrong actions: they are not-to-be-done. So realists will reject any theory of properties that treats wrongness and BASness equivalently.^8,9 Similar arguments can be constructed, mutatis mutandis, for predicate and mereological nominalism. There is also a more particular reason moral realists should not accept predicate nominalism. According to moral realism, objects fall under moral predicates because they possess moral properties. According to predicate nominalism, objects possess moral properties because they fall under moral predicates. Combining these views generates explanatory circularity.

That leaves resemblance nominalism, transcendent realism, immanent realism and trope theory.¹⁰ All these views take possession of properties to mark genuine similarities between objects, which the moral realist will want to extend to the case of wrongness. However, moral realists seeking a companion here should choose carefully. One issue involves causation. Several of these theories of properties give a key role to causation in determining which similarities are genuine, and hence which properties exist.¹¹ According to the immanent realist D. M. Armstrong (1997b), for example, we have reason to postulate universals in so far as their instances are involved in true singular causal statements. For example, it is the fire’s hotness, not the fire itself, that causes the water to boil. Relatedly, D. H. Mellor (1997) has argued that the strongest case for the existence of universals concerns causation: singular causation requires the truth of causal laws, which in turn quantify over properties. Which properties exist, on these views, is a matter of a posteriori investigation, with the properties winning an existence ticket being all and only those that feature in the laws of nature thus discovered. Armstrong and Mellor argue, further, that we cannot assume that the properties of everyday experience, such as blueness, are genuine, since there may be many different types of property (distinguished by causal powers) whose instances cause sensations of blue in ordinary observers. Trope theorists also typically take which tropes there are to be closely tied to causation. A standard argument in favour of tropes, for instance, is that they are required to be the relata in causal statements: the hotness of the fire causes the water to boil, where this hotness is at once particular and nature-furnishing, that is, a trope. ‘Tropes are the levers of the world’ (Schaffer 2001, p. 248). Resemblance nominalism has fewer direct connections to the notion of causation, but is still a risky partner for some moral realists. Resemblance nominalists typically take resemblance facts to be primitive and definitive only of the so-called ‘sparse’ or natural properties, so BASness is excluded (for example, Rodriguez-Pereyra 2002, pp. 50–1, 62–5). Resemblance nominalism is therefore an uneasy partner for moral realists who seek to deny the naturalness of moral properties, whether or not that notion is directly tied to causation.

Amongst immanent realists, trope theorists and resemblance nominalists, therefore, there is a common view that genuine properties are those whose instances can be causally efficacious or are otherwise natural. Applying such views to the moral case suggests that if moral realists accept any of these theories of properties, at least as traditionally presented, they are under pressure to accept that moral properties can have causally efficacious instances, or are otherwise natural. In short, these theories of properties best fit with naturalistic versions of moral realism (for example, Railton 1986).

For realists who deny the causal efficacy of moral properties—that is, non-naturalists—but who (being realists) accept that moral predicates have a property-ascribing function, there are two remaining options. The first is to adopt non-standard versions of resemblance nominalism, immanent realism or trope theory that reject the connection between genuine similarity and causation or naturalness. For realists who take this path one task will be to offer an alternative account of the type of similarity that makes for property-hood, one that allows for causally inefficacious yet genuine properties.¹² There seem several options for non-naturalist realists here. They could argue, for instance, that genuine similarities are those that we need to deploy in explanation, where this is not limited to the causal case.¹³ Or they could argue that genuine similarities (hence properties) are those that we need to recognize for any intellectually indispensable project, where such projects include practical deliberation as well as (scientific) explanation (for example, Enoch 2011; FitzPatrick 2019).

The final option for non-naturalist realists is transcendental realism. At least as traditionally conceived, transcendental realism does not limit properties to those whose instances can be causally efficacious, so within this framework moral non-naturalists can take moral properties to be similarity-defining yet non-causal. However, it seems likely that the dialectic just rehearsed will also play out for this combination of views. This is because transcendental realists who are moral realists will want to claim that not every predicate that correctly applies to some object corresponds to a universal (consider ‘grue’ and ‘BAS’).¹⁴ So they too will need to offer a general account of which similarities are genuine and which are not, and non-naturalism rules out appealing to causation. Either way, therefore, moral non-naturalists require an account of genuine similarity that goes beyond the causal and naturalist accounts popular amongst property realists.

To sum up this section, I have argued for two claims. First, that moral realism is in theoretical tension with predicate, class and mereological nominalism. Second, that if non-naturalist moral realists adopt resemblance nominalism, immanent realism, transcendental realism or trope theory, they had best adopt non-standard versions of these views that reject the connection between genuine similarity and causation or naturalness.

IV

In this section, I consider connections between moral realism and some standard arguments for the existence of properties. It is noteworthy that in both ontology and metaethics, one type of argument is semantic: we need to postulate the existence of properties in order to make sense of our language (for example, Russell 1912, ch. 9; Shafer-Landau 2003, p. 24; Brink 1989, pp. 25–9). I argue that the standard semantic arguments for the existence of properties do not support moral realism.

According to the reference argument, there are some sentences that are true and involve terms that apparently refer to properties. Hence properties exist (Armstrong 1978, vol. I, pp. 58–63; 1997a, pp. 105–8; Lewis 1997b, pp. 194–6). Consider, for instance, predicates, and the simple sentence ‘a is F’. A tempting semantic analysis, revealing ontological commitment, is as follows:

(s) ‘a is F’ is true if and only if there is an x such that ‘a’ refers to x and there is a Y such that ‘F’ refers to Y, and Y is possessed by x.

On this view, predicates refer to properties. Hence, accepting that some sentences correctly apply predicates, we are committed to the existence of properties.

A second group of terms that involves apparent reference to properties is that of abstract singular terms. The following examples are common:

• (1) Humility is a virtue.

• (2) Redness is a sign of ripeness.

The truth of claims such as (1) and (2) seemingly commits us to entities corresponding to the terms ‘humility’ and ‘redness’, that is, to properties (Armstrong 1997a).

A related argument involves true sentences that apparently quantify over (rather than refer to) properties. Examples include:

• (3) He has the same virtues as his father.

• (4) There are undiscovered fundamental physical properties.

According to the quantification argument, there are some sentences—such as (3) and (4)—that are true and that apparently quantify over properties. So properties exist (Armstrong 1997a; Edwards 2014, pp. 6–7).

The reference and quantification arguments are not without critics. One type of response is to suggest paraphrases of the relevant sentences which preserve their meaning but eschew the apparent commitment to properties. A suitable paraphrase of (2), for instance, might be ‘Red fruits are ripe’ (Devitt 1997; Armstrong 1997a). A worry with paraphrasing responses, however, is that they are ad hoc. Even assuming that a suitable paraphrase can be found for these particular examples, there is no guarantee that they are always available.

David Lewis extends this criticism to offer a synthesis of the reference and quantification arguments. Considering sentences such as (1) and (2), he says:

Prima facie, these sentences contain names that cannot be taken to denote particular, individual things. What is the semantic role of these words? If we are to do compositional semantics in the way that is best developed, we need entities to assign as semantic values to these words, entities that will encode their semantic roles. Perhaps sometimes we might find paraphrases that will absolve us from the need to subject the original sentence to semantic analysis … But even if such paraphrases sometimes exist—even if they always exist, which seems unlikely—they work piecemeal and frustrate any systematic approach to semantics. (Lewis 1997b, pp. 194-5)

By a ‘compositional’ semantics Lewis means an assignment of meanings to sentences that can demonstrate how those meanings are built up systematically from the meanings of their constituent parts. Suppose we accept that such a semantics works best when it assigns entities to predicates like ‘red’ and abstract singular terms like ‘redness’, and that it works systematically when it assigns such entities to all meaningful predicates and abstract singular terms. What type of entities should it assign? Should it, for example, assign universals (the realist answer)? Or classes of particulars (the class nominalist answer)? Lewis prefers the latter because the former would break the connection between universals and genuine similarity. A dramatic example is provided by:

• (5) Grueness does not make for resemblance among all its instances (Lewis 1997b, p. 195).

This sentence is true, and, given compositional semantics, there is reason to assign an entity as the semantic value of ‘grueness’. Yet if we took this entity to be a universal, we would be forced to deny one of the essential features of universals: the marking of genuine similarities. Best then, to take the entities assigned to abstract singular terms (and predicates) to be classes of particulars. This is Lewis’s conclusion. As far as compositional semantics is concerned, predicates and abstract singular terms are set-contributing rather than universal-ascribing (Lewis 1997b, p. 195; Edwards 2014, pp. 82–4).

The upshot is that the semantic argument supports class nominalism rather than any more substantial theory of properties. We saw in the previous section that class nominalism is inconsistent (or at least in tension) with moral realism. Therefore the semantic argument for the existence of properties does not support moral realism.¹⁵

V

This completes my partial investigation of the connections between theories of properties and morality. In this section, I consider the second of the two questions distinguished in §i: how should we respond to the connections thus established?

A natural principle is to start from the general and move to the particular. On this view we should first articulate and defend a general theory of properties, for example, and then apply it to the particular case of morality, forming our metaethical views accordingly. For example, if our theory of properties entails that properties are necessarily causal, then we should reject metaethical theories that posit the existence of causally inert moral properties (that is, non-naturalist realism). The risk of this strategy is one of over-generalization. This is the vice of forming a general theory of a phenomenon based on a limited range of its instances, and then taking that theory to determine our views in every other case. I think that sometimes this risk is manifest in moving from theories of properties to theories of morality. It is notable, for instance, how those formulating theories of properties take as their paradigms properties as they appear in perception or scientific explanation. It is perhaps not surprising if theories based on generalization from such cases rule out the possibility of autonomous, causally inert, moral properties.

Suikkanen (2024, §v) argues that trope theory is theoretically friendly towards moral non-naturalism in so far as it provides a framework in which the latter can explain moral supervenience. According to that framework, moral properties are properties of properties, and properties are sets of resembling tropes. So, for example, consider the maximal set, S₁, containing all and only being-a-hurtful-lie tropes {T₁, T₂, …, T_n}. Suppose that all hurtful lies are wrong. Suikkanen’s trope theorist understands this situation as one where a wrongness trope, W₁, is compresent with S₁, where ‘compresence’ is the tie that bundles tropes together to form objects. This in turn provides an explanation of moral supervenience, as follows. Because the moral properties of a particular object are determined by the moral properties of the (non-moral) types it belongs to (such as the type: being a hurtful lie), any two particular objects that share all the same (non-moral) types will share the same moral properties. Hence there can be no moral difference without non-moral difference.

I am unsure whether trope theory is as helpful for non-naturalists as Suikkanen suggests. First, as he acknowledges, traditional versions of trope theory take tropes to be located in space and time (perhaps even individuated thus) and the compresence relation is understood in terms of spatio-temporal co-location (Williams 1997, pp. 115–17; Schaffer 2001). They also identify tropes with causal relata.¹⁶ Obviously, qua non-naturalist, Suikkanen’s trope theorist cannot accept this last view. But they seem hard pressed, too, to accept the first two, given that they take (some) tropes to be compresent with sets of tropes, and sets are typically taken to be abstract, that is, not spatio-temporally located.¹⁷ Second, it is unclear how, on trope theory, the moral properties of particular objects can be determined by the properties of the kinds to which they belong. For example, consider S₁ (the maximal set of being-a-hurtful-lie tropes), which is compresent with a particular wrongness trope, W₁. Now consider the particular lie that Jon told last Thursday, and the being-a-hurtful-lie trope it contains as a part, call this T₁. T₁ is a member of S₁. Jon’s action is also wrong, so assume (for the time being at least), in line with the general trope theory of possessing a property, that this entails that T₁ is compresent with a distinct wrongness trope, W₂. The question now is why it should be the case that S₁ being compresent with W₁, and T₁ being a member of S₁, jointly determines T₁ being compresent with W₂. S₁, W₁, T₁ and W₂ are, after all, distinct, particular existents. Admittedly, the foregoing picture, which takes particular objects such as Jon’s lie to be compresent with wrongness tropes (like W₂), is not one that Suikkanen explicitly attributes to the trope theorist (here my previous assumption lapses). Suikkanen describes the view as holding that ‘the primary bearers of normative properties are kinds’ (2024, §v, my emphasis). One might then ask how we are to interpret the implication that particulars like Jon’s lie are secondary bearers of normative properties. One might say (and this I take to be the view Suikkanen attributes to the trope theorist) that for Jon’s lie to be wrong just is for T₁ to be a member of S₁ and for S₁ to be compresent with W₁. But in that case, a particular action’s being wrong is not a matter of the tropes that constitute it including a wrongness trope—there is no wrongness trope that is compresent with the bundle of tropes that constitutes Jon’s action—which seems to abandon altogether the thought that objects possess moral properties in the same way that they possess non-moral properties such as being hurtful.¹⁸ A third worry is that we lack an explanation of why higher-order moral tropes are compresent only with maximal sets of resembling tropes (that is, with properties). Consider, for example, the non-maximal set S₂, which contains all of the being-a-hurtful-lie tropes except T₁. Suppose that S₂, rather than S₁, is compresent with wrongness trope W₁. Then, accepting that the moral properties of particulars (that is, bundles of tropes) derive from the higher-order moral properties of the sets of tropes of which they contain a member, all of the actions that include being-hurtful-lie tropes are wrong, except Jon’s lie. And, moreover, this is compatible with Jon’s lie being exactly qualitatively identical with some other lies: a moral difference without a non-moral difference. It is unclear why such a situation could not arise, according to trope theory.¹⁹

Putting these worries aside, though, Suikkanen’s paper (§vi) draws a particular conclusion from the premiss that only the trope theory of properties is compatible with both naturalist and non-naturalist moral realism. The conclusion is that we should accept trope theory.

The general principle behind this argument seems to be as follows:

Positive Epistemic Neutrality (pen) . Other things being equal, we should accept theories in area A that uniquely do not foreclose otherwise live options in related area B.

In the case at hand, we are assuming that both realist non-naturalism and naturalism are—independent of the debate about properties—live options. Only trope theory does not determine the choice between them. Therefore, we should accept trope theory (other things being equal).

I am sceptical that pen is true. It appears to take something that is partly an artefact of our current state of knowledge (that a theory is neutral between theories that we cannot yet decide between) to be a reason to think that a theory is true. Ultimately, truth will not be neutral between B₁ and B₂, so it is unclear why the neutrality of A₁ counts in favour of its truth. What might be the rationale for such a principle? Suikkanen’s autonomy assumption (§i) may be one:

[W]hich side in this debate [between naturalism and non-naturalism] is right ought to turn only on what is distinctive of the normative properties in particular … [This] entails that the metaethical debate should not be decided based on considerations external to the debate, for example, concerning the nature of properties generally. (Suikkanen 2024, p. 115)

I am sceptical. Which side of the debate between naturalism and non-naturalism is true certainly ought to turn on the nature of the properties posited by each side, but it may also turn on whether those proposed natures are consistent (or coherent) with any well-established theory of properties. With such a theory in hand, metaethical debates could well be decided, in part, by issues external to the moral case, issues such as the general nature of properties. Of course, this can only happen when we have a well-established general theory of properties. The risk of over-generalization is the risk that our actual theories might not be well-established. But we should not rule out the possibility that such theories, were they to be available, determine metaethical issues. This possibility seems to be part of many methodologies in metaethics (for example, Timmons 1999, ch. 1), and extends not just to general theories of properties but those of meaning and knowledge as well.

An alternative to pen is:

Negative Epistemic Neutrality (nen). Other things being equal, we should not accept theories in area A that foreclose otherwise live options in related area B.

nen avoids the objection to pen since it does not treat neutrality as a reason to think that a theory is true. nen seems especially plausible where area A is general (for instance, the theory of properties) and area B is one species that falls under it (for instance, the theory of moral properties). This is in part because of the vice of over-generalization: there is a risk, for example, that theory A₁ only forecloses otherwise live options in B because the subject matter of B has not been considered in arguing for A₁. Of course, if A₁ is well-established (that is, one has a sufficient argument in favour of it), then other things are not equal and we should accept it, even if it forecloses what would otherwise be live options in B. In philosophical debates, however, most often other things are equal in this sense: few theories are well-established.

If we replace pen with nen, the imperative to accept trope theory does not follow. Granting that (class) nominalism, immanent realism and transcendent realism foreclose otherwise live options in metaethics, nen entails that we should not accept those theories (other things being equal). But from the fact that trope theory uniquely does not foreclose otherwise live options, it does not follow that we should accept it. Another option is agnosticism.²⁰ On this view, in our current state of knowledge, it is less important to consider which views to accept as final truth, and more important to consider which views hang together. Accepting Suikkanen’s arguments, non-naturalism hangs best with transcendental realism or trope theory. Naturalism hangs best with immanent realism or trope theory. Methodologically, we can rest here, accepting (for the time being) no theory of properties, and no theory of moral properties, but instead merely recognizing the connections between them. What hangs together need not come down to earth.²¹

References

Footnotes