Introduction

The Aristotelian view that ordinary objects are composites of form and matter (“hylomorphism”) has seen a resurgence of popularity, e.g. Fine (1999, 2003, 2008), Johnston (2002, 2006), and Koslicki (2006, 2008).1 Prima facie, a hylomorphic view of ordinary objects has an advantage over matter-only views of ordinary objects, such as e.g. Jubien (2001), Quine (1960), or Sider (2001), when it comes to accounting for the modal properties of ordinary objects.2 By literally building form into what it is to be the object, hylomorphists have a starting point for modality which matter-only adherents do not. Given hylomorphism, objects are inseparable from (because constituted by) their form — and, by extension, from the kind/modal profile/modal properties which follow from their having the form they do.3 Prima facie, for the hylomorphist, accounting for an ordinary object’s modal properties is easy in a way it isn’t for adherents of matter-only views.

One might worry, however, that this prima facie ease is an illusion. On a matter-only view the hard question is modal: which modal profile does that (statue-shaped) object have? Does it have the modal profile of a statue, a lump, a mere aggregate? On a hylomorphic view the hard question is ontological: which objects exist? Does a statue (matter-m + statue-form), a lump (matter-m + lump-form), and/or a mere aggregate (matter-m + mere aggregate-form) exist? Koslicki appeals to science: our best science tells us which kinds of objects exist and philosophy defers (2008, 200–234). Fine appeals to a principle of modal plenitude: every form that’s eligible to combine with matter m does so: a statue, a lump, and a mere aggregate all coincide (1999). Johnston appeals to genuine parts and principles of unity: every combination of genuine parts which is bound by a principle of unity is an object (2006, 658).

My aim in this paper is to defend a novel answer to the hard question for hylomorphism. In section I, I discuss the material aspect of ordinary objects. In section II, I discuss the formal aspect of ordinary objects. In section III, I defend response-dependent hylomorphism. I argue that which forms combine with which bits of matter — and, thus, which ordinary objects exist — depends not only on the intrinsic properties and situation of the matter and form, but, also, on how subjects respond to relevant bits of matter. In section IV, I argue that a central advantage of response-dependent hylomorphism over non-response-dependent hylomorphism is the former allows us to tell a more satisfactory modal grounding story.

I. The Material Part of Ordinary Objects

One strategy of the hylomorphist who wishes to avoid primitive modality is to separate the modal aspect of ordinary objects from the empirical aspect of ordinary objects and then to tell a separate grounding story for each. I’ll follow this strategy. In particular, I’ll argue that ordinary objects are composites of a material part which lacks (non-trivial) de re modality and a formal part which gives rise to the (non-trivial) de re modal properties the ordinary object has.4

An object’s de re modal properties are the properties it has which are prefixed by necessity or possibility, e.g. being necessarily self-identical, being necessarily mammalian, being possibly red, being possibly taller than 6’.5 There is no universally accepted demarcation between trivial modal properties and non-trivial modal properties. In general, however, modal properties which rely on the specific nature of the object which has them and which are had by some objects and lacked by others (e.g. being necessarily mammalian, having part p necessarily, being possibly taller than 6’, being possibly in France) tend to be considered non-trivial. Whereas modal properties which do not rely on the specific nature of the object which has them and which are had by all objects (e.g. being necessarily mammalian or non-mammalian, being necessarily self-identical, being possibly red if actually red) tend to be considered trivial. This general distinction isn’t perfect.6 However, it’s sufficient for the purpose at hand.

Call entities which lack (non-trivial) modal properties “n-entities”. The claim that there are n-entities is controversial.7 There are some who, tacitly or overtly, defend their existence, e.g. Quine, Gibbard, Lewis, and Sidelle.8 There are many others who find n-entities unpalatable. The intuitive reason to endorse n-entities is the belief that some entities exist coupled with the belief that the world doesn’t have enough modal structure to support the existence of objects which have (non-trivial) modal properties.9 Two aspects of n-entities will be important in what follows: (i) grounding modality worries do not arise for n-entities, i.e. because they have no (non-trivial) de re modal properties, and (ii) n-entities have empirical properties. An object’s empirical properties are the properties it has which aren’t prefixed by necessity or possibility, e.g. being self-identical, being mammalian, being red, being taller than 6′.10 I’ve defended the existence of n-entities elsewhere.11 Since my central aim here is to show how hylomorphists can answer the hard question without relying on primitive (non-trivial) de re modality and a variety of accounts of the material part are compatible with this aim, I won’t repeat my arguments in favor of taking the material part to be an n-entity.

The general hylomorphist strategy I defend should be of interest even to those who reject n-entities. Anyone who thinks the modal properties of ordinary objects differ from the modal properties of the basic fundamental entities can utilize the general grounding strategy I present. N-entities are a place-holder for what our best science tells us there is. Arguably, whatever this turns out to be, it’s not going to have exactly the modal properties we associate with ordinary objects. A grounding story, like mine, which tells how to get from the existence of such basic entities to ordinary objects should, thus, be of general interest even to those who reject the specifics of the view I endorse, i.e. who reject the existence of n-entities.12

II. The Formal Part of Ordinary Objects

The formal part of an ordinary object is a sort property. A sort property is a property which individuates an object. It tells what sort of object the object is and gives the object’s modal nature. Being a rock, being a dog, and being a tree are all sort properties because they’re all properties which individuate sorts of objects: rocks, dogs, and trees. Being blue, being big, and being waterish, on the other hand, aren’t sort properties. They’re all properties objects might have, but they aren’t properties which pick out a certain sort of object.13

Three aspects of sort properties will be important in what follows: (i) sort properties specify the empirical properties an n-entity must have if it’s apt to be an object of the relevant sort, (ii) entailment relations between sort properties and other properties make it the case that an object which has a sort property as a part thereby instantiates the modal properties entailed by the sort property, and (iii) sort properties exist prior to, and independently of, their instances.

With regard to (i), consider, for example, the sort property being a rock. The sort property being a rock specifies the empirical properties an n-entity must have to be rock-apt. Although it’s geologists, rather than philosophers, that focus specifically on the empirical properties of rocks, we have — simply qua human beings that interact with rocks frequently — a fairly good idea of the relevant empirical properties. These include such properties as having a mass between m (larger than pebble mass) and mn (smaller than boulder mass), being made of granite or limestone or quartz or …, being appropriately distinct from one’s environment (i.e. not embedded in a mountain made of the the same substance), etc.

With regard to (ii), consider, for example, the sort property being a rock. This property stands in entailment and compatibility relations with other properties, e.g. being a rock entails being material, being cohesive, and is compatible with being painted white, being first at location l and then at location l’. It is because of these entailment and compatibility relations that rocks have the (non-trivial) modal properties they have, e.g. being necessarily material, being necessarily cohesive, being possibly painted white, being possibly in the next room.14

With regard to (iii), in section III I will argue that ordinary objects (e.g. rocks) reduce to n-entities and sort-properties. In order for such a reduction to be possible, sort-properties must exist independently of, and prior to, the ordinary objects which have them as parts. Certain accounts of properties, such as an Aristotelian one in which the existence of a universal depends on the existence of an instance of it, are thus incompatible with my project.

There are further questions to explore with regard to sort properties, e.g. what overall account of properties do they fall within, are they abundant or sparse, how general can they be (e.g. is being a mammal a sort property or is it only properties denoting specific species of mammals — being a dog, being a horse — which are sort properties), for each sort s exactly what modal properties follow from an object’s being of sort s. However, just as with n-entities, the exact details don’t matter. Any account of properties which is compatible with sort properties playing the three roles outlined above is acceptable. I will, hence, leave this question for another day.15

III. Response-Dependent Hylomorphism

Response-dependent hylomorphists argue that not just any matter (n-entity) and any form (sort-property) combine to make an ordinary object of sort s. Rather, n-entity n sums with sort-property S to form an object of sort s (at spacetime region r) only if: (1) n is s-apt, and (2) some subject is having the s-response to n. For example, n-entity 1 sums with being a rock to form a rock only if: (1) n-entity 1 is rock-apt, and (2) some subject is having the rock-response to n-entity 1. Understanding these conditions requires understanding what it is for an n-entity to be s-apt and what it is for a subject to have the s-response to an n-entity.

S-Aptness (of n-entities)

S-Aptness: N-entity n is s-apt to be an object of sort s iff none of the empirical properties that n has rule out its being an object of sort s.

Consider an n-entity, n1, which is round, grey, made of granite, has a mass of 500g, and is distinct from its surroundings. N1 is rock-apt. There are lots of other objects n1 is, also, s-apt to be. For instance, it’s mere-aggregate-apt and time-slice-of-a-rock-apt. Just as there are lots of objects n1 is s-apt to be, there are lots of objects n1 isn’t s-apt to be. For instance, n1 isn’t elephant-apt (it’s lacking the empirical property standing in evolutionary chain e) and n1 isn’t swing-set-apt (it’s lacking the empirical property being swing set shaped).

As mentioned in section II, it’s the sort properties themselves which ground the fact that n1 is rock-apt, but not swing-set-apt. The sort property being a rock determines what empirical properties are compatible with being a rock. The sort property being a swing set determines what empirical properties are compatible with being a swing set. It’s the fact that the empirical properties n1 has are compatible with being a rock (as determined by the sort property being a rock) that makes it the case that n1 is rock-apt. It’s the fact that the empirical properties n1 has aren’t compatible with being a swing set (as determined by the sort property being a swing set) that makes it the case that n1 isn’t swing-set-apt.

Sort-Responses (to n-entities)

Sort-Responses: To respond to an n-entity as if it’s an object of sort s.

To have a sort-response to n-entity n, is (tacitly or overtly) to respond to n as if it’s an object of a particular sort, i.e. an object which has a modal nature. To have the rock-response to n1 is to treat n1 as if it’s a rock; to think of or behave toward n1 as if it’s a rock. The exact nature of human sort-responses is a question for psychologists, not for philosophers. First-person experience suggests that we have such sort-responses automatically and that our having sort-responses is, generally, a passive process rather than an active process. We don’t contemplate some matter, think about the various sorts we could categorize it into, and then choose one. Rather, we simply have the s-response to an n-entity. Which sort-responses we have is, presumably, influenced by the culture we live in and the language we speak, as well as by our biology.16,17

Not just any response is a sort-response. Sort-responses pick out sorts of objects, i.e. objects which have a modal nature. The rock-response, dog-response, and tree-response are sort-responses because they’re responses that pick out objects with modal natures: rocks, dogs, and trees. The blue-response, big-response, and rock-like-thing-over-there-response, on the other hand, aren’t sort-response. They’re all responses one might have upon encountering an n-entity, but they don’t pick out a sort of object.

Any subject which has the requisite cognitive abilities can have sort-responses.18 Humans have sort-responses. Whether non-human animals have sort-responses is a question for ecologists. The philosopher can certainly imagine alien subjects who would have different sort-responses than we have. Imagine a species which believes in in-cars and out-cars rather than in cars. Or a species of mereological essentialists who believe in objects which can’t lose parts. The objects they create won’t, of course, be that different from the ones we create. Aliens with different sort-responses can create mere aggregates where, faced with qualitatively identical n-entities, we’d create rocks. But they can’t create elephants where we’d create rocks. The empirical properties of the n-entities limit the subject’s creative power.

Response-Dependent Composition

When the two conditions are met — i.e. when (1) there’s an s-apt n-entity n at region r and, (2) a subject has the s-response to n — an ordinary object of sort s comes into existence. For example, when (1) there’s a rock-apt n-entity at region r and (2) a subject has the rock rock-response to the rock-apt n-entity, then a rock comes into existence. Response-dependence, thus, comes into the picture at the level of composition. An s-apt n-entity (“matter”) exists and sort property s (“form”) exists. What doesn’t exist, prior to the response of a subject, is the sum of the n-entity and the sort property. Composition (between an n-entity and a sort property) is, thus, response-dependent.

Ordinary Object Composition (RD): An n-entity n which is s-apt sums with the sort property being s to compose an ordinary object of sort s iff a subject has the s-response to n.

For example, an n-entity sums with the sort property being a rock to compose a rock iff the n-entity is rock-apt and a subject has the rock-response to it. Likewise, an n-entity sums with the sort property being a tree to compose a tree iff it is tree-apt and a subject has the tree-response to it.19,20 Ordinary objects are mereological sums composed of n-entities and sort properties, so e.g.

Rock R = rock-apt n-entity + being a rock

Tree T = tree-apt n-entity + being a tree21

It is worth noting that ordinary object composition (RD) concerns the existence of ordinary objects, not their modal properties. A response-dependent hylomorphism which is response-dependent due to endorsing ordinary object composition (RD) is as realist about modality as is any non-response-dependent view. So, for instance, although the advocate of ordinary object composition (RD) thinks that ordinary objects depend on subjects for their existence, she doesn’t think they depend on subjects for their modal profiles. It’s an ordinary object’s sort-property (e.g. being a rock) which determines its modal profile. If we’d been different (i.e. lacked rock-responses) rocks wouldn’t have existed, but it would still be the case that, were rocks to exist, they would be necessarily cohesive and possibly grey.

There are stronger and weaker forms of response-dependent composition depending, for example, on (i) whether sort-responses are needed only to trigger the composition relation or both to trigger and maintain it, (ii) whether subjects have to be in direct causal contact22 with the n-entity they’re responding to or can respond indirectly, and (iii) who’s responsible for triggering/maintaining the composition relation — e.g. any subject, a privileged subject or group of subjects, an ideal responder, etc. The particular view I favor is a strongly response-dependent one in which a subject is required both to trigger and maintain the composition relation, subjects have to be in direct causal contact with the n-entity they’re responding to, and the responses of any subject who’s cognitively sophisticated enough to make the relevant sort distinctions are sufficient to create an object.23

Although a full defense of strong response-dependence is beyond the scope of this paper (and I’m far less committed to strong response-dependence than I am to the general claim that the best way to account for the existence and modal properties of ordinary objects is via hylomorphism), a few words regarding the prima facie plausibility of strong response-dependence are in order. A consequence of endorsing a strongly response-dependent form of hylomorphism about ordinary objects is that, absent responding subjects, there are no ordinary objects such as rocks, trees, and dogs. One might take this consequence to be so costly — Surely the rock continues to exist even when no one is looking! Surely there were dinosaurs (even though there were no responding subjects in the Jurassic Age)! — as to outweigh any benefits of a strongly response-dependent hylomorphism.

The advocate of a strongly response-dependent hylomorphism has two options here: error-theory or linguistic revision. The underlying ontology is the same in both cases. There are rock-apt n-entities and there were (in the Jurassic Age) dinosaur-apt n-entities. That there are such n-entities has nothing to do with subjects; it has only to do with what the sort properties being a rock and being a dinosaur require and with whether there are any n-entities that meet these requirements. There is a rock-apt n-entity in my backyard. Sometimes a subject is responding to it — at these times there’s a rock. Other times no subject is responding to it. At this point, error theory and linguistic revision diverge.

The error theorist says the word “rock” refers only to objects which instantiate the property being a rock. The only objects which instantiate being a rock are hylomorphic objects which have a rock-apt n-entity as their matter and the sort property being a rock as their form. Since — given strong response-dependence — there is no such object when there is no responding subject, the error theorist says there is no rock (although, of course, there is still a rock-apt n-entity). The error theorist, thus, thinks some of the claims we make about ordinary objects — “There were dinosaurs.” “Rocks can exist unresponded to.” — are false. The error theorist attempts to mitigate this cost by pointing out that which ordinary object claims are false is systematic and explicable. In particular, we’ll get the existence and persistence conditions of ordinary objects wrong when we confuse the existence/persistence of an s-apt n-entity for the existence/persistence of an ordinary object of sort s. We think dinosaurs existed because we mistake the existence of dinosaur-apt n-entities for the existence of dinosaurs. We think rocks exist unresponded to because we mistake the existence of rock-apt n-entities for the existence of rocks. Similar ‘errors’ occur whenever natural language hasn’t caught up with (or can’t be bothered to note in everyday discourse) scientific discoveries. We say, “The sun rises in the East” even though we haven’t believed Ptolemaic astronomy in some time. We say, “That train is moving and this one is standing still” even though we haven’t believed in Newtonian physics for some time. The philosophical view I’m proposing — a strongly response-dependent hylomorphism about ordinary objects — is, like Copernican astronomy and Einsteinian physics, an ontological view about the nature of reality. It’s a new view. We don’t yet think of ordinary objects as being sums of n-entities and sort-properties and we don’t yet think of ordinary objects as depending on subjects’ responses (to apt n-entities) for their existence. So, of course, we don’t speak as if they are so dependent. Of course, we systematically mistake rock-apt n-entities for rocks. That de re modality isn’t a deep feature of the world — and, hence, that objects having modal properties aren’t as independent/fundamental/real as we once thought they were — isn’t something reality wears on its sleeve.

Some may find error theory unappealing. The linguistic revisionist says, if strong response-dependence is correct, the best thing isn’t to endorse an error theory, it’s to revise the extension of our ordinary object terms. Let “rock” denote both rock-sums (i.e. sums of rock-apt n-entities and the sort property being a rock — what the error theorist calls “rocks”) and rock-apt n-entities. Let “dinosaur” denote both dinosaur-sums (i.e. sums of dinosaur-apt n-entities and the sort property being a dinosaur — what the error theorist calls “dinosaurs”) and dinosaur-apt n-entities. Now we can say that there were dinosaurs and that rocks exist unresponded to. (But we can no longer ‘read’ an objects’ modal profile off its being denoted by a certain English word, e.g. some rocks — the ones that are rock-apt n-entities — won’t have any of the modal properties we intuitively think rocks have, other rocks — the ones that are rock-sums — will have many of these properties.)24

I won’t take a stand here on whether the adherent of a strongly response-dependent hylomorphism should prefer error theory or linguistic revision. The adherent of any ontological theory which deviates from our folk ontology has, to some degree, to be either an error theorist or a linguistic revisionist. N-entities aren’t part of our folk ontology, so the adherent of a strongly response-dependent hylomorphism will have, to some degree, to be either an error theorist or a linguistic revisionist. This is a (relatively benign, in my opinion) feature of the view. It is shocking to find out there were no dinosaurs or that rocks exist unresponded to. But once one understands why this is the case (i.e. because of the underlying ontology of ordinary objects), what the motivation is for taking ordinary objects to have such ontology (i.e. to account for their modal properties in a world that, independent of subjects, lacks modal umpf), and that there’s a near-by close contender which has fooled us into thinking dinosaurs existed and rocks exist unresponded to (i.e. dinosaur-apt n-entities and rock-apt n-entities) this initial shock is no longer compelling.25

IV. Hylomorphism, Response-Dependence, and Grounding Modality

Hylomorphists argue ordinary objects consist of a material part and a formal part. The formal part is standardly taken to give the object’s modal profile. Hylomorphic objects, thus, come ready made with their modal natures. For hylomorphists, essence is prior to existence. Hylomorphists, thus, approach grounding modality differently than do non-hylomorphists.26 Rather than taking the existence of some object o for granted and then telling a grounding story about which modal properties o has, hylomorphists take the modal profiles of objects for granted and then tell a grounding story about which objects exist. For hylomorphists, the hard part isn’t grounding modality; the hard part is grounding existence.

All hylomorphists use form to (easily) ground modality. An ordinary object is the sort of object it is in virtue of having a certain sort property as a part, e.g. R is a rock in virtue of having the sort property being a rock as its formal part. An ordinary object has the modal properties it has in virtue of the sort of object it is, e.g. R has the (non-trivial) de re modal properties it has in virtue of being a rock.

All hylomorphists then face the hard question: In virtue of what does rock R (as opposed to only lump L or aggregate-of-granite-atoms A) exist?27 Here the response-dependent hylomorphist has an advantage over the advocate of non-response-dependent hylomorphism. According to response-dependent hylomorphism, the existence of ordinary objects is jointly grounded by the existence and empirical properties of n-entities, by the existence of sort properties, and by the responses of subjects (which determine whether an s-apt n-entity sums with sort property S to compose an ordinary object of sort s).28 Non-response dependent hylomorphists have no such tidy grounding story to tell.2930

Conclusion

I have suggested that, although the hard question for matter-only accounts of ordinary objects is, “Which modal profile does that object have?”, the hard question for hylomorphic accounts of ordinary objects is, “Which objects exist?”. Matter-only view have an easy time grounding the existence of objects which have modal properties and a hard time grounding the object’s having of modal properties. Hylomorphic views have an easy time grounding an object’s having of modal properties and a hard time grounding the existence of objects which have modal properties. I have argued that response-dependent hylomorphists have an easier time answering their hard grounding question than do either matter-only adherents or non-response-dependent hylomorphists and I have presented a sketch of my preferred version of response-dependent hylomorphism.31