Our everyday experiences present us with a wide array of objects: dogsand cats, tables and chairs, trees and their branches, and so forth.These sorts of ordinary objects may seem fairly unproblematic incomparison to entities like numbers, propositions, tropes, holes,points of space, and moments of time. Yet, on closer inspection, theyare at least as puzzling, if not more so.
This entry concerns a variety of problems that arise in connectionwith ordinary material objects. Proposed solutions to the problems, aswe shall see, almost invariably have one of three unpalatableconsequences. First, overpopulating the world with things that seemnot to be there, be it too many tables, or too many things in oneplace, or too many causes of the same events, or a plenitude ofobjects with extraordinary mereological or modal profiles. Second,unpalatable arbitrariness, be it arbitrariness concerning which kindsof objects there are, which objects do and don’t belong to agiven kind, which modal profiles are instantiated by which objects, orwhich objects together compose a further object. Third, unpalatableindeterminacies, be it indeterminate truth values, indeterminateidentities, ontic indeterminacy, or existential indeterminacy. Forthis reason, some view the problems as constituting a powerful casefor theelimination of the ordinary objects that give rise tothem in the first place.
Section 1 articulates a variety of ways of departing from an ordinary,conservative conception of which objects there are, either byeliminating ordinary objects or by permitting more objects than wewould ordinarily take to exist. Section 2 examines the puzzles andarguments that are meant to motivate these departures. Section 3examines some arguments against eliminative and permissive views.Finally, Section 4, turns from the question of which objects exist tothe question of which objects existfundamentally.
We find ourselves naturally inclined to make certain judgments aboutwhich objects are before us in various situations. Looking at a pooltable just before the break, we are naturally inclined to judge thereto be sixteen pool balls on the table, perhaps various parts of theindividual balls (their top and bottom halves), and no othermacroscopic objects. Looking at my nightstand, I am naturally inclinedto judge there to be an alarm clock, a lamp, their various parts(lampshade, buttons, cords), and nothing else.
Conservative views are those according to which these sorts ofjudgments are by and large correct. Giving a precise characterizationof conservatism, or of ordinary objects, is no easy task. Veryroughly, ordinary objects are objects belonging to kinds that we arenaturally inclined to regard as having instances on the basis of ourperceptual experiences:dog,tree,table,and so forth. Extraordinary objects, by contrast, are macroscopicobjects belonging to kinds that we are not ordinarily inclined toregard as having instances. (More on these in§1.3.) And conservatism is roughly the view that there are just the ordinaryobjects and none of the extraordinary objects.[1]
Revisionary views about which objects there are are those that departin one way or another from conservatism. These include botheliminative views, on which there are fewer ordinary objects than arerecognized by conservatives, and permissive views, on which there areextraordinary objects that conservatives do not recognize. There is,however, some controversy about whether these departures fromconservatism actually deserve to be called ‘revisionary’.As we shall see in§3.1, many eliminativists and permissivists take their views to be entirelycompatible with common sense and ordinary belief.
Our target question—namely, which macroscopic objectsexist—may be distinguished from related but independentquestions concerning thenature of ordinary objects. Someviews about the natures of objects may seem to be at odds with commonsense, for instance, the view that ordinary objects can’tsurvive the loss of any of their parts, or that ordinary objects areall mind-dependent. But these views are entirely compatible withconservatism, as characterized above, because they do not (or at leastneed not) have any revisionary implications regardingwhichobjects there are at a given place and time. That said, questionsabout the nature of ordinary objects are intimately connected withquestions about which objects exist, insofar as certain views aboutthe nature of these objects (including those just mentioned) providethe resources for addressing some of the puzzles and arguments thatmotivate revisionary conceptions.
A few terminological preliminaries. ‘Object’ is used inits narrow sense, which applies only to individual material objectsand not to other sorts of entities like numbers or events.‘Part’ is used in its ordinary sense, on which it is nottrue—or at least not trivially true—that things are partsof themselves. And when the entry notes that some objects composesomething, or that they have a fusion, what is meant is that there issomething that has each of them as parts and every part of whichoverlaps at least one of them. The entry avoids the word‘sum’, but newcomers to this literature should bear inmind that it gets used in different ways: ‘sum of x and y’is sometimes used to meanobject that is composed of x and yand is other times used to meanobject that is composed of x and yand that has all of its parts essentially.[2]
Eliminative views are those that deny the existence of some wide rangeof ordinary objects. (Denying merely that ordinary objects arefundamental is not by itself enough to qualify as an eliminativist;see§4 below.)
Some eliminativists accept nihilism, the thesis that no objects evercompose anything. In other words, every object is mereologicallysimple (i.e., partless). Together with the plausible assumption thatordinary objects (if they exist) are all composite objects, nihilismentails that there are no ordinary objects. Nihilists typically acceptthat there are countless microscopic objects: although there are“simples arranged dogwise” and “simples arrangedstatuewise”, there are no dogs or statues. But nihilism is alsocompatible with existence monism—the thesis that there is asingle, all-encompassing simple (the cosmos, a.k.a. “theblobject”)—as well as the extreme nihilist thesis thatthere are no objects whatsoever.[3]
Since many of the arguments for eliminativism actually fall short ofestablishing that composition never occurs, it is also open toeliminativists to reject nihilism and accept certain classes ofcomposites. Many eliminativists make an exception for persons andother organisms. Some, for instance, accept organicism, the thesisthat some objects compose something just in case the activities ofthose objects constitute a life. In other words, organisms are theonly composite objects.[4]
The motivations for making an exception for organisms vary. VanInwagen (1990: Ch. 12) accepts organicism on the grounds that ityields the best answer to the special composition question(“under what conditions do some objects composesomething?”), one that allows for one’s own existence andphysicality, while at the same time escaping various problems thatarise for competing answers. Merricks (2001: Ch. 4) argues thatpersons and some other composites must be recognized on account oftheir nonredundant causal powers. Making such exceptions naturallygives rise to concerns about the stability of the resulting positions,either because the reasoning behind allowing the exceptions threatensto generalize to all ordinary objects or because the arguments foreliminating ordinary objects threaten to generalize to the objects onewishes to permit.[5]
It is also open to eliminativists to adopt non-nihilistic views thatare fairly liberal about composition, allowing that composition occursat least as often as we ordinarily suppose (if not more so). PeterUnger is one such non-nihilistic eliminativist:
There is nothing in these arguments [for eliminativism] to deny theidea, common enough, that there are physical objects with a diametergreater than four feet and less than five. Indeed, the exhibited[arguments] allow us still to maintain that there are physical objectsof a variety of shapes and sizes, and with various particular spatialrelations and velocities with respect to each other. It is simply thatno such objects will be ordinary things; none are stones or planets orpieces of furniture. (1979b: 150)
While Unger does use the label ‘nihilism’ for his view, heis not a nihilist in our sense because he affirms that there is ahighly visible composite object occupying the exact location where wetake the table to be. He is, however, an eliminativist insofar as hedenies that that object is a table.[6]
Permissive views are those according to which there are wide swathesof extraordinary objects.
Universalism is the permissivist thesis that composition isunrestricted: for any objects, there is a single object that iscomposed of those objects. What universalism does not tell us is whichkinds of objects there are. Whenever there are some atomsarranged turkeywise, universalism entails that there issomeobject that they compose. But it remains open to universalists (likethe aforementioned non-nihilistic eliminativists) to deny that thiscomposite is a turkey. However, assuming that there are such objectsas turkeys, trout, and their front and back halves, universalism willentail that there are trout-turkeys, where a trout-turkey is a singleobject composed of the undetached front half of a trout and theundetached back half of a turkey. These are objects that have bothfins and feathers and whose finned parts may be a good distance fromtheir feathered parts.[7]
Diachronic universalism is the permissivist thesis that, for any timesand any function from those times to sets of objects that exist atthose times, there is an object that exists at just those times andhas exactly those parts at those times. Roughly: there is an objectcorresponding to every filled region of spacetime. So, assuming thatyour kitchen table and living room table both exist, there also existsa klable: an object that’s entirely made up of your kitchentable every day from midnight till noon and is entirely made up ofyour living room table from noon till midnight. This is an objectthat, twice a day, instantly and imperceptibly shifts its location.[8]
Some accept even more plenitudinous forms of permissivism.Formulations vary, but the rough idea is that, so long as theempirical facts don’t rule out there being an object with agiven modal property in a given location, then there is an object inthat location with that modal property. When there is a red car parkedin the garage, the empirical facts (e.g., that there is nothing bluethere) do rule out there being an object in the exact location of thecar that is necessarily blue. But they do not rule out there being anobject colocated with the car that is inside the garage as a matter ofmetaphysical necessity (an “incar”) and that would ceaseto exist if the car were to leave the garage.[9]
One further way of being a permissivist is by permitting a multitudeofparts of ordinary objects that we do not naturally judgethem to have. For instance, one might hold that, in addition toordinary parts like arms and legs, you have extraordinary parts likeleg complements, where your left-leg complement is an object made upof all of you except for your left leg. Together with some naturalassumptions (e.g., about regions of space), leg complements and alegion of other extraordinary parts will be delivered by the doctrineof arbitrary undetached parts—or DAUP—the thesis that forany material object o, if r is the region of space occupied by o, andif r′ is an occupiable sub-region of r, then there exists amaterial object that exactly occupies r′ and which is part of o.(Roughly: for every region of space within the boundaries of a given aobject, that object has a part that exactly fills that region.)[10]
Sorites arguments proceed from a premise to the effect that minutedifferences cannot make a difference with respect to whether someproperty F (or kind K) is instantiated to the conclusion that nothing(or everything) is F (or a K). Here is a sorites argument for theelimination of stones:
Premises SR2 and SR3 together entail that, for any finite number ofatoms, nothing made up of that many atoms is a stone. But this,together with SR1, entails that there are no stones.[11]
Similar arguments may be given for the elimination of individualordinary objects. One can construct a sorites series of contiguousbits of matter, running from a bit of matter, mk, at thepeak of Kilimanjaro to a bit of matter, mp, in thesurrounding plains. From the sorites premise that a bit of matterthat’s n inches along the path from mp tomk is part of Kilimanjaro iff a bit of matter that’sn+1 inches along the path is part of Kilimanjaro (for any number n),together with the fact that mp is not part of Kilimanjaro,we reach the absurd conclusion that mk isn’t part ofKilimanjaro. So, by reductio, we may conclude that Kilimanjaro doesnot exist.[12]
Why accept SR3? Imagine a series of cases, beginning with a caseinvolving a single atom and terminating with a case involving whatwould seem to be a paradigm stone, where each case differs from thepreceding case only by the addition of a single atom. It seems highlyimplausible that there should be adjacent cases in any such serieswhere there is a stone in one case but not in the other. Rejecting SR3would look to commit one to just such a sharp cut-off.
But one can deny that SR3 is true without accepting that there is asharp transition from stones to non-stones in such series, that is,without accepting that there is some specific object in the seriesthat definitely is a stone and whose successor definitely is not astone. For one may instead hold that there is a range of cases inwhich it is vague whether the object in question is a stone.[13]
Here is an illustration of how that sort of strategy might go. Let Sbe some object in the series that clearly seems to be a stone, let NSbe an object that clearly seems to be a non-stone, and let BS be anobject that seems to be a borderline case of being a stone. One mightsuggest that ‘stone’ is vague as a result of there being arange of candidate precise meanings (or“precisifications”) for the word ‘stone’,
‘S is a stone’ is true because S falls under all of theseprecisifications of ‘stone’. ‘NS is a stone’is false because NS doesn’t fall under any of them. And‘BS is a stone’ is neither true nor false because BS fallsunder some but not all of the precisifications. And then SR3 itselfturns out to be false: onevery precisification of‘stone’, there is some object in the series such that itbut not its successor falls under that precisification. (This issometimes known as a “supervaluationist” account.)
Defenders of sorites arguments often complain that this line ofresponse still commits one tosome “sharp statustransition”, for instance, a sharp transition from a case inwhich ‘there is a stone’ is true to a case in which it isneither true nor false.[14]
It is natural to suppose that objects sometimes do, and other times donot, compose a further object. When a hammer head is firmly affixed toa handle, they compose something, namely, a hammer. When they’reon opposite ends of the room, they don’t compose anything. Thefollowing argument—known as “the argument fromvagueness”—purports to show that this natural assumptionis mistaken.
If the argument is sound then either universalism or nihilism must becorrect, though which of them is correct would have to be decided onindependent grounds.[15]
A sorites series for composition is a series of cases running from acase in which composition doesn’t occur to a case in whichcomposition does occur, where adjacent cases are extremely similar inall of the respects that one would ordinarily take to be relevant towhether composition occurs (e.g., the spatial and causal relationsamong the objects in question). Understood in this way, AV1 should beunobjectionable. If it’s true that the handle and head composesomething only once the hammer is assembled, then a moment-by-momentseries of cases running from the beginning to the end of the assemblyof the hammer would be just such a series. Premise AV2 looks trivial:any such series obviously must containsome transition fromcomposition not occurring to composition occurring, and there eitherwill or won’t be a determinate fact of the matter about whereexactly that transition occurs in any given series.
AV3 is plausible as well. If composition occurs in one case but not inanother, then surely there must be some explanation for why that is.In other words, the facts about composition are not“brute”. Yet the sorts of differences that one finds amongadjacent cases in a sorites series for composition—for instance,that the handle and head are a fraction of a centimeter closertogether in the one than in the other—can’t plausiblyexplain why composition occurs in one case but not in the other.[16]
Certain sorts of eliminativists are well positioned to resist AV3without having to accept that compositional facts are brute. Suppose,for instance, that one accepts a view on which conscious beings arethe only composite objects. Such eliminativists will deny that thereis a sorites series for composition running from the beginning to theend of the assembly process, since they will deny that anything iscomposed of the handle and head (or that thereare a handleand head) even at the end of the series. Every sorites series forcomposition, by their lights, will have to run from a case involving aconscious being to one that doesn’t. And assuming that thatthere can’t be borderline cases of consciousness, every suchseries will contain a sharp cut-off with respect to the presence ofthe additional subject of consciousness. This, in turn, is poised toexplain why composition occurs in the one case but not the other.[17]
Why, though, should anyone accept AV4? On the face of it, it seemsjust as clear that there can be borderline cases of composition (e.g.,when the hammer head is just beginning to be affixed to the handle) asit is that there can be borderline cases of redness and baldness. Thisis not, however, “just another sorites,” to be blocked inwhichever way one blocks the sorites arguments in §2.1.That’s because questions about when composition occurs look tobe intimately bound up with questions about how many objects exist.This suggests the following line of argument in defense of AV4, noanalogue of which is available for other sorts of sorites arguments.[18]
To see the motivation for AV6, notice that if the handle and head docompose something then there are three things: the handle, the head,and a hammer. If they don’t, then there are only two things: thehandle and head. And if it is vague whether they do, then it will bevague whether there are two things or three. As for AV7, notice thatone can specify how many objects there are using what would seem to beentirely precise vocabulary. For any finite number, one can produce a“numerical sentence” saying that there are exactly thatmany concrete objects. Here, for instance, is the numerical sentencefor two: ‘∃x∃y(x≠y & Cx & Cy &∀z(Cz→(x=z ∨ y=z)))’. (The restriction toconcreta ensures that numerical sentences aren’t trivially falsesimply on account of there being infinitely many numbers, sets, and soforth.) And since these numerical sentences contain no vaguevocabulary, it would seem to follow that it cannot be indeterminatehow many objects there are.
AV6 can be resisted by denying that composition affects the number ofobjects in the way suggested. For instance, one might contend thateven before the handle and head definitely come to compose something,there exists an object—a“proto-hammer”—located in the region that the two ofthem jointly occupy. The proto-hammer definitely exists, but it is aborderline case of composition: it is indeterminate whether the handleand head compose the proto-hammer or whether they instead composenothing at all (in which case the proto-hammer has no parts).[19]
Alternatively, one might resist AV7 by pinning the vagueness on thequantifiers in the numerical sentence. After all, what seems to bevague is whether the handle and head areeverything thatthere is and whetherthere issomething other thanthe handle and head. But it is difficult to see how the quantifierscan be vague and, in particular, how their vagueness could beaccounted for on the sort of standard, precisificational account ofvagueness discussed in§2.1.[20]
Ordinary objects are constituted by, or made up out of, aggregates ofmatter. A gold ring is constituted by a certain piece of gold. Claystatues are constituted by pieces of clay. We are naturally inclinedto regard the statue and the piece of clay as being one and the sameobject, an object that simply belongs to multiple kinds(statue andpiece of clay).
The puzzles of material constitution put pressure on this naturalinclination. Here is one such puzzle. Let ‘Athena’ name acertain clay statue and let ‘Piece’ name the piece of claythat constitutes it. What is puzzling is that all of the followingseem true:
The motivation behind MC2 is that Athena seemingly has exactly thesame location and exactly the same parts as Piece. So‘Athena’ and ‘Piece’ are plausibly justdifferent names for the same thing. The motivation behind MC3 is thatPiece and Athena seem to have different modal properties: Piece isable to survive being flattened and Athena isn’t. MC4 followsfrom the Principle of the Indiscernibility of Identicals (a.k.a.Leibniz’s Law): ∀x∀y(x=y → ∀P(Px iffPy)). In other words, if x and y are identical, then they had betterhave all the same properties. After all, if they are identical, thenthere is onlyone thing there to have or lack any given property.[21]
The puzzles are sometimes taken to motivate eliminativism, sinceeliminativists can simply deny MC1: there are no statues (and perhapsno pieces of clay either).[22]
More often, however, the puzzles are taken to motivate constitutionalpluralism, the thesis that ordinary objects are typically, if notalways, distinct from the aggregates of matter that constitute them.(‘Typically’ because, in rare cases in which the ordinaryobject and the aggregate come into existence at the same time andcease to exist at the same time, some pluralists will take theordinary object to be identical to the aggregate.) Pluralists rejectMC2: clay statues are not identical to the pieces of clay thatconstitute them. Pluralists may deny that having the same parts at agiven time suffices for identity, or they may instead deny that thestatue and the piece of clay have all of the same parts.[23]
One of the main problems facing the pluralist solution is thegrounding problem: the modal differences between Athena and Piece(e.g., that the one but not the other can survive being flattened)seem to stand in need of explanation and yet there seems to be nofurther difference between them that is poised to explain, or ground,these differences.[24]
Defenders of the pluralist response to these puzzles may, by similarreasoning, be led to accept that, in special cases, two objects of thesame kind can coincide. Suppose, for instance, that we have afantastically big net (Thin) with very thin netting. We then roll itup into a long rope, and we weave that rope into a smaller net (Thick)with a thicker weave. Since the nets intuitively have different modalproperties—Thin, but not Thick, can survive the unraveling ofthe thicker net—the same sort of reasoning that leads one toreject MC2 underwrites an argument that the nets are not identical. Inother words, there are two exactly colocated objects, both of whichare nets.[25]
Constitutional monists, according to whom Athena is identical toPiece, will deny MC3. There are various ways of developing the monistresponse. First, one might insist that both Athena and Piece (which,on this view, are identical) can survive flattening: upon flattening,Athena ceases to be a statue, but does not cease to exist. We can callthose who opt for this approach ‘phasalists’, since theytakebeing a statue to be a temporary phase that Piece (i.e.,Athena) is passing through.[26]
Alternatively, monists might deny that Piece (i.e., Athena) cansurvive being flattened. When Piece is flattened, Piece ceases toexist, at which point an entirely new piece of clay (composed of thesame atoms) comes into existence. This strategy is sometimes called“the doctrine of dominant kinds”, since the idea is thatwhen an object belongs to multiple kinds, the object has thepersistence conditions associated with whichever of the kinds“dominates” the others. Becausestatue dominatespiece, Piece is a statue essentially, and therefore cannotsurvive ceasing to be statue-shaped.[27]
Finally, monists might agree that Piece is able to survive beingflattened and that Athena is not able to survive being flattened, andyet deny that Athena and Piece have different properties. How can thatbe? On one version of this approach (often associated with counterparttheory), the idea is that ‘is able to survive beingflattened’ is context-sensitive, expressing one property whenaffixed to ‘Athena’ and another when affixed to‘Piece’. On another, the idea is that ‘is able tosurvive being flattened’ does not express a property at all.Either way, we do not end up with any one property that Piece has butAthena lacks.[28]
A wooden ship is constructed and christened ‘Theseus II’.As planks come loose over the years, they are discarded and replaced.After three hundred years, the last of the original planks isreplaced. Call the resulting ship ‘the mended ship’. Thedescendants of the original owners have been collecting the discardedoriginal planks, and—three hundred years after thechristening—they obtain the last of them and construct a shipthat is indistinguishable from the original. Call the resulting ship‘the reconstructed ship’. Which, if either, of these twoships is identical to Theseus II? It is natural to suppose that thereis no fact of the matter: it is indeterminate which of the two shipsis Theseus II.
However, there arguably can never be indeterminate cases ofidentity:
ST2 relies on two seemingly innocuous inferences: (a) from its beingindeterminate whether Theseus II is identical to the mended ship toTheseus II’s being indeterminately identical to the mended shipand (b) from there to Theseus II’s having the property of beingindeterminately identical to the mended ship. ST3 seems trivial aswell: the mended ship is definitely self-identical, so it does notitself have this property. ST4 looks to be an immediate consequence ofLeibniz’s Law: if Theseus II and the mended ship differ withrespect to even one property, then they are distinct. ST5 is trivial:if they’re not identical, then it isn’t indeterminatewhether they are identical.[29]
Eliminativists may go on to argue from ST6 to the conclusion thatthere are no ships, as follows. If indeed it isn’t indeterminatewhether Theseus II is the mended ship, then there would seem to befive options:
Options (i) and (ii) seem intolerably arbitrary, since the mended shipand reconstructed ship seem to have equal claim to being Theseus II.Option (iii) is out as well. If Theseus II is identical to both ships,then (by the transitivity of identity) they must be identical to oneanother; but they cannot be identical because they have differentproperties (e.g., the one but not the other is composed of theoriginal planks). Option (iv) is problematic as well. The history ofmaintenance by itself would have sufficed for the persistence ofTheseus II; the preservation and reassembly of the original parts byitself would likewise have sufficed for the persistence of Theseus II;and here we have managed to secure both. As Parfit (1971: 5) wouldsay, “How could a double success be a failure?” Thus, weget (v) from argument by elimination. Fitting.[30]
Some respond to the argument by maintaining that it is indeterminatewhich of various objects ‘Theseus II’ picks out. Ifthat’s right, then ST2 is arguably false. One cannot infer theexistence of an individual who is indeterminately identical to Suefrom the fact that it is indeterminate whether Sue (or rather Morgan)is Harry’s best friend. Analogously, one cannot infer theexistence of an individual that is indeterminately identical to themended ship from the fact that it is indeterminate whether Theseus IIis the mended ship.[31]
The prima facie problem with this response is that there do not seemto be multiple objects such that it’s indeterminate which one‘Theseus II’ picks out. After all, when ‘TheseusII’ was first introduced, there was only one ship around toreceive the name! One can address this problem by maintaining that,despite appearances, two ships were present at the christening: onethat would later be composed of entirely different planks and anotherthat would later be reassembled from a pile of discarded planks. Whatis indeterminate is which of these two temporarily colocated ships waschristened ‘Theseus II’.[32]
Other responses are available. One might deny ST2 on the grounds thatthere is no such property as the property of being indeterminatelyidentical to Theseus II. One might deny ST3, affirming that the mendedship is indeterminately identical to the mended ship. One might denyST4 by denying that the distinctness of Theseus II and the mended shipcan be inferred from the fact that they do not share the indicatedproperty. Or one might embrace option (iv), on the (stage-theoretic)grounds that the ship that existed at the time of the christening isnot identical to any ship that exists at any earlier or later time.[33]
Arguments from arbitrariness turn on the observation that there wouldseem to be no ontologically significant difference between certainordinary and extraordinary objects, that is, no difference betweenthem that can account for why there would be things of the one kindbut not the other. Here is an example (drawn from Hawthorne 2006:vii):
The idea behind AR2 is that islands and incars (see§1.3) would seem to be objects of broadly the same sort, namely, objectsthat go out of existence simply by virtue of changing theirorientation with respect to some other thing (the water level in theone case, the garage in the other), without their constitutive matterundergoing any intrinsic change. The idea behind AR3 is that, if theretruly are islands but no incars, then this is something that wouldseem to stand in need of explanation: there would have to be somethingin virtue of which it’s the case. To think otherwise would be totake the facts about what exists to be arbitrary in a way that theyplausibly are not.
Similar arguments may be used to establish the existence of legcomplements (on the grounds that there’s no ontologicallysignificant difference between them and legs) and trout-turkeys (onthe grounds that there’s no ontologically significant differencebetween them and scattered objects like solar systems).[34]
Eliminativists may of course resist the argument by denying AR1.[35]
The argument may also be resisted by denying AR2 and identifying someontologically significant difference between islands and incars. Forinstance, a certain sort of anti-realist will say that which objectsthere are is largely determined by which objects we take there to be.Accordingly, the very fact that we take there to be islands but notincars marks an ontologically significant difference between them.Alternatively, one may attempt to identify an ontologicallysignificant difference between the ordinary and extraordinary objectswithout endorsing anti-realism. In the case at hand, one might resistAR2 by insisting that islands have importantly different persistenceconditions from incars. Incars are meant to cease to exist when theirmatter ceases to be inside a garage. But islands, contra hypothesis,do not cease to exist when they are completely submerged; they merelycease to be islands.[36]
How about AR3? Part of why it seems arbitrary to countenance islandsbut not incars is that one would seem to be privileging islands overincars by virtue of taking them to exist. For this reason, proponentsof certain deflationary ontological views are well positioned to denyAR3. Relativists, for instance, may maintain that islands exist andincars do not exist—relative to our conceptual scheme, that is.Relative to other, equally good schemes, incars exist and islands donot. Quantifier variantists, who maintain that there are counterpartsof our quantifiers that are on a par with ours and that range overthings that do not exist—but rather exist*—may maintainthat islands exist but do not exist* while incars exist* but do notexist. On such views, islands and incars receive a uniform treatmentat bottom; islands are not getting any sort of “specialtreatment” that cries out for explanation.[37]
We encounter some atoms arranged treewise and some atoms arrangeddogwise, and we naturally take there to be a dog and a tree. But thereare different ways we might have carved up such a situation intoobjects. Instead of taking there to be a tree there, we might insteadhave taken there to be a trog: a partly furry, partly wooden objectcomposed of the dog and the tree-trunk.
Why, though, do we naturally take there to trees rather than trogs?Plausibly, this is largely the result of various biological andcultural contingencies. If so, then there arguably is little reason toexpect that our beliefs about which objects there are would be evenapproximately correct. This realization, in turn, is meant to debunkour beliefs about which objects there are:
The idea behind DK1 is that we are inclined to believe in trees ratherthan trogs largely because prevailing conventions in the communitieswe were born into generally prohibit treating some things as the partsof a single object unless they are connected or in some other wayunified. These conventions themselves likely trace back to an innatetendency to perceive only certain arrays of qualities as being borneby a single object and its being adaptive for creatures like us to soperceive the world. But the facts about which distributions of atomsdo compose something, or about which arrays of qualities truly areborne by a single object, have no role to play in explaining why thisis adaptive. Thus, it would seem that we divide up the world intoobjects the way that we do for reasons having nothing at all to dowith how the world actually is divided up.
The idea behind DK2 is that if there truly is this sort of disconnectbetween the object facts and the factors that lead us to our objectbeliefs, then it could only be a lucky coincidence if those factorsled us to beliefs that lined up with the object facts. And the ideabehind DK3 is that since we have no rational grounds for believingthat we got lucky, we shouldn’t believe that we did, in whichcase we should suspend our beliefs about which objects there are and,in particular, our belief that there are trees.[38]
Such debunking arguments fall short of establishing that eliminativismis true or that conservatism is false. But, if successful, they dolend powerful support to eliminativism, by effectively neutralizingany reasons we might take ourselves to have for accepting conservatismor for wanting to resist the arguments for eliminativism.[39]
The arguments also lend indirect support to permissivism, insofar aspermissivists are well positioned to deny DK2. By permissivist lights,having accurate beliefs about which kinds of objects there are is atrivial accomplishment (not a coincidence), since there are objectsanswering to virtually every way that we might have perceptually andconceptually divided up a situation into objects. The ordinary andextraordinary objects are all already out there waiting to be noticed;all that our conventions do is determine which ones are selected for attention.[40]
Deflationists also seem well positioned to deny DK2. Relativists willsay that, while we could easily have divided up the world differently,we could not easily had divided up the world incorrectly. For had wedivided the world into trogs rather than trees, we would then have hada different conceptual scheme, and we would have correctly believedthat trogs exist-relative-to-that-scheme. Quantifier variantists willsay that, had we divided the world into trogs but not trees, we wouldthen have correctly believed that trogs exist*.[41]
Alternatively, one might try to resist DK1 by identifying anexplanatory connection between the way the world is divided up and ourbeliefs about how it is divided up. For instance, one might say thatwe have the object beliefs that we do as a result of intelligentdesign: God, wanting us to have largely accurate beliefs, arranged forus to have experiences that represent trees and not trogs. Or onemight take a rationalist line, according to which, through somecapacity for rational insight, we intellectually apprehend relevantfacts about which objects together compose something. Or one might optfor an anti-realist line and insist that there is a mind-to-worldexplanatory connection: object beliefs determine the object facts andare therefore an excellent guide to which kinds exist.[42]
Overdetermination arguments aim to establish that ordinary objects ofvarious kinds do not exist, by way of showing that there is nodistinctive causal work for them to do. Here is one such argument:
For the purposes of this argument, ‘atoms’ may beunderstood as a placeholder for whichever microscopic objects or stufffeature in the best microphysical explanations of observable reality.These may turn out to be the composite atoms of chemistry, they may bemereological simples, or they may even be a nonparticulate“quantum froth”.
One could resist OD1 by maintaining that some things that are causedby baseballs are not also caused by their atoms. On one way ofdeveloping this line of response, baseballs “trump” theiratoms: atoms arranged baseballwise can’t collectively causeanything to happen so long as they’re parts of the baseball. Onanother, there is a division of causal labor: baseballs cause eventsinvolving macroscopic items like the shattering of windows, whiletheir atoms cause events involving microscopic items like thescatterings of atoms arranged windowwise. Both strategies, however,look to be in tension with the plausible claim that there is acomplete causal explanation for every physical event wholly in termsof microphysical items. Moreover, this line of response would seem torequire that baseballs have emergent properties—causallyefficacious properties that cannot be accounted for in terms of theproperties of their atomic parts—which seems implausible.[44]
OD2 can be motivated as follows:
Let us say that an event e is overdetermined by o1 ando2 just in case:
This can be taken as a stipulation about how‘overdetermined’ is to be understood in the argument, thuspreempting nebulous debates about whether satisfying these fiveconditions suffices for “real” or “genuine”overdetermination. To say that o1 is causally relevant too2’s causing e is to say that o1 entersinto the explanation of how o2 causes e to occur in one ofthe following ways: by causing o2 to cause e, by beingcaused by o2 to cause e, by jointly causing e together witho2, or—where o2 is a plurality ofobjects—by being one of them.[45]
Can OD6 be resisted? The idea would have to be that, although someevents are caused both by atoms and by baseballs composed of thoseatoms, those events are not overdetermined (in the indicated sense).But if they are not overdetermined, then which of the five conditionsfor overdetermination do the baseball and the atoms fail to meet? Thisline of response takes for granted that (i) and (ii) are satisfied.And it is extremely plausible that (iii) and (iv) would be satisfiedas well. However it is that baseballs “get in on theaction”, it isn’t by entering into the causal explanationofhow the atoms manage to cause things. Baseballsdon’t cause their atoms to shatter windows, nor do their atomscause them to shatter windows. So those who would deny OD6 will needto deny that condition (v) is satisfied, by taking the baseball to beidentical to the atoms. See§3.3 below for discussion of the thesis that objects are identical totheir various parts.
Why accept OD7? In certain cases, overdetermination strikes us as anovert violation of Ockham’s Razor: do not multiply entitiesbeyond necessity. But given the intimate connection between baseballsand their atoms, it is natural to feel that even if these do count ascases of overdetermination (in the indicated sense), this isn’tan especially objectionable sort of overdetermination. One may thenattempt to resist OD7 by articulating a further condition whichdistinguishes problematic from unproblematic cases ofoverdetermination. For instance, one might hold that overdeterminationby o1 and o2 is unproblematic so long aso1 and o2 aren’t entirely independent.[46]
Even supposing, however, that the line between objectionable andunobjectionable sorts of overdetermination can be drawn in somesatisfactory way, there would still be pressure to accept OD7. Weshould accept that something other than the atoms shatters the windowonly if we have good reason to believe in this something. But there isnoexplanatory need to posit baseballs, since there is acomplete causal explanation for all of the relevant events wholly interms of the activities of the atoms. And the debunking arguments in§2.6 purport to show that our ordinary perceptual reasons for believing inbaseballs are no good. So we would seem to have no good reason at allto accept that there are baseballs, in which case we ought to accept OD7.[47]
Premise OD4 can be motivated in much the same way as OD7. If baseballsdon’t cause anything to happen, then we have no good reason tobelieve in them, in which case we should accept OD4. One might alsogive a more direct defense of OD4 by appealing to the controversialEleatic Principle (a.k.a. Alexander’s Dictum), according towhich everything that exists has causal powers. Together with theplausible assumption that if baseballs don’t cause anythingit’s because theycan’t cause anything, theEleatic Principle entails OD4.[48]
The office appears to contain a single wooden desk. The desk isconstituted by a single hunk of wood whose surface forms a sharpboundary with the environment, without even a single cellulosemolecule coming loose from the others. Call this hunk of wood‘Woodrow’. Now consider the object consisting of all ofWoodrow’s parts except for a single cellulose molecule,‘Molly’, making up part of Woodrow’s surface. Callthis ever-so-slightly smaller hunk of wood‘Woodrow-minus’. Because Woodrow-minus is extraordinarilysimilar to Woodrow, there is considerable pressure to accept thatWoodrow-minus is a desk as well. This, in short, is the problem of themany.
PM1 and PM2 straightforwardly entail that PM3 is false; one of theseclaims has to go.[49]
PM1 is plausible. Woodrow-minus seems to have everything that it takesto be a desk: it looks like a desk, it’s shaped like a desk,it’s got a flat writing surface, and so forth. Accordingly, itseems arbitrary to suppose that Woodrow but not Woodrow-minus is adesk. Moreover, if Molly were removed, Woodrow-minus would certainlythen be a desk. But since Woodrow-minus doesn’t itself undergoany interesting change when Molly is removed (after all, Mollyisn’t evenpart of Woodrow-minus), it stands to reasonthat Woodrow-minus must likewise be a desk even while Molly isattached to it.
One might deny PM1 on the grounds that being a desk is a“maximal” property, that is, a property of an object thatcannot be shared by large parts of that object. Since Woodrow is adesk, and since Woodrow-minus is a large part of Woodrow,Woodrow-minus is not a desk.[50]
But this style of response can be rendered unavailable by introducingan element of vagueness into our story. Suppose now that Molly hasbegun to come loose from the other molecules, in such a way that it isnaturally described as being a borderline part of the desk in theoffice. Let Woodrow-plus be the aggregate of cellulose molecules thatdefinitely has Molly as a part. PM1 can then be replaced withPM1′:
Woodrow-plus and Woodrow-minus each seem to have everything that ittakes to be a desk, and neither seems to be a better candidate thanthe other for being a desk. PM2 would then be replaced withPM2′:
PM1′ can be resisted by proponents of the supervaluationiststrategy sketched in§2.1 above. The vague term ‘desk’ has multipleprecisifications, some of which apply to Woodrow-plus, some of whichapply to Woodrow-minus, butnone of which applies to both.Accordingly, PM1′ is false on some precisifications, andtherefore is not true simpliciter.[52]
Constitutional pluralists can deny both PM2 and PM2′. Regardingthe original story, they may insist that neither Woodrow norWoodrow-minus is a desk. Each is a mere hunk of wood, and no mere hunkof wood is a desk. Rather, there is exactly one desk, it isconstituted by Woodrow, and while Woodrow-minus would constitute thatdesk if Molly were removed, as things stand it constitutes nothing atall. Regarding the revised story, pluralists may again say that thereis exactly one desk, neither Woodrow-plus nor Woodrow-minus is a desk,and it is simply indeterminate whether it is Woodrow-plus orWoodrow-minus that constitutes that desk. In that case, PM2′ isfalse. It’s true that each is a desk iff the otheris—since neither is a desk—but it doesn’t followthat there’s more than one or fewer than one desk.[53]
Finally, one might deny PM3, either by accepting an eliminative viewon which there is no desk in the office or by accepting a permissiveview on which there is more than one desk in the office. Proponents ofthe latter response will end up committed to far more than two desks,however. By parity of reason, there will also be a desk composed ofall of the cellulose molecules except Nelly (≠ Molly). Likewise forOllie. And so on. So there will be at least as many desks as there arecellulose molecules on the surface of the desk.[54]
Universalism seems to conflict with our intuitive judgment that thefront halves of trout and the back halves of turkeys do not composeanything. Put another way, universalism seems to be open to fairlyobvious counterexamples. Here is an argument from counterexamplesagainst universalism:
Similar arguments may be lodged against other revisionary theses. Thevarious forms of eliminativism wrongly imply that there are nostatues; plenitudinism wrongly implies that there are incars; thedoctrine of arbitrary undetached parts wrongly implies that there areleg complements; and so forth.[55]
Compatibilist accounts of the apparent counterexamples take thetargeted revisionary views to be entirely compatible with theintuitions or beliefs that are meant to motivate CX2. Such accountsoften take the form of assimilating recalcitrant ordinary utterancesto some familiar linguistic phenomenon that is known to be potentiallymisleading. For instance, when an ordinary speaker looks in the fridgeand says ‘there’s no beer’, she obviouslydoesn’t mean to be saying that there is no beer anywhere in theuniverse. Rather, she is tacitly restricting her quantifier to thingsthat are in the fridge. Universalists often suggest that somethingsimilar is going on when ordinary speakers say ‘there are notrout-turkeys’ (or ‘there’s nothing that has bothfins and feathers’). Speakers are tacitly restricting theirquantifiers to ordinary objects, and what they are saying is entirelycompatible with there beingnon-ordinary finned-and-featheredthings like trout-turkeys. Universalists may then hold that theargument from counterexamples rests on an equivocation. If thequantifiers are meant to be restricted to ordinary objects, then CX2is true, but CX1 is false: universalism does not entail that anyordinary things are trout-turkeys. If on the other hand thequantifiers are meant to be entirely unrestricted then CX2 is false;but in denying CX2 (so understood), one is not running afoul ofanything we are inclined to say or believe or intuit.[56]
This is just one of many compatibilist strategies that have beendeployed in defense of revisionary views. Universalists have alsoinvoked an ambiguity in ‘object’ to explain the appeal of‘there is no object that has both fins and feathers’.Eliminativists have claimed that ordinary utterances of ‘thereare statues’ are instances of “loose talk”, or thatthey are context-sensitive, or that quantifiers are being used in aspecial technical sense in the “ontology room”.[57]
One common complaint about compatibilist accounts is that theseproposals about what we are saying and what we believe arelinguistically or psychologically implausible. For instance, whenordinary speakers are speaking loosely or restricting theirquantifiers, they will typically balk when their remarks are taken atface value. (“There’s no beer anywhere in theworld?”) But this sort of evidence seems just to be missing inthe cases at hand. (“You think there literally arestatues?” “There’s nothingat all with bothfins and feathers?”) Others have criticized compatibiliststrategies by pointing to limitations of their recipes forparaphrasing ordinary utterances. For instance, Uzquiano (2004:434–435) argues that the standard eliminativist strategy ofparaphrasing constructions of the form ‘there are Fs’ interms of there being atoms arranged Fwise cannot be adapted to handleconstructions like ‘some Fs are touching each other’.[58]
Revisionaries may instead wish to give incompatibilist accounts of theputative counterexamples, conceding that the revisionary views theydefend are incompatible with ordinary belief (ordinary discourse,common sense, intuition, etc.), but maintaining that the mistakes canbe explained or excused. For instance, revisionaries may contend thatthe mistaken beliefs are nevertheless justified, so long as one is notaware of the defeaters that undercut our usual justification (e.g.,those mentioned in§2.6). Or they may contend that ordinary speakers are not especiallycommitted to these beliefs, which may in turn suggest that they do notdeserve to be treated as data for purposes of philosophical inquiry.Or they may call attention to some respect in which the ordinaryutterances and beliefs are “nearly as good as true”.[59]
One way of approaching the question of whether there are statues is byasking whether the correct interpretation of the English language isone according to which ordinary utterances of ‘there arestatues’ come out true. The interpretation of populations ofspeakers is plausibly governed by a principle of charity thatprohibits the gratuitous ascription of false beliefs and utterances topopulations of speakers. Such a principle—which is independentlymotivated by reflection on how it is that utterances come to have themeanings that they do—can be put to work in arguments for theexistence of ordinary objects and for the nonexistence ofextraordinary objects. Here is one such argument from charity:
To see the idea behind CH1, notice that both eliminativists andconservatives can agree that there are atoms arranged statuewise. Thequestion is whether the English sentence ‘there arestatues’ should be interpreted in such a way that the existenceof such atoms suffices for it to come out true. Let us callinterpretations of ‘there are statues’ on which theexistence of atoms arranged statuewise suffices for its truthliberal, and interpretations on which that does not sufficefor its truthdemanding. The idea then is that, given theavailability of both liberal and demanding interpretations, the formerwould clearly be more charitable. CH2 is motivated by the thought thatthere are no other content-determining factors that favor a demandinginterpretation over a liberal interpretation, in which case charitywins out and ‘there are statues’ is true. CH3 looks to bea straightforward application of a plausible disquotation principle:if sentence S says that p, and S is true, then p.[60]
One might challenge CH1 on the grounds that charitable interpretationis a holistic matter, and, while the liberal interpretations arecharitable in some respects, they are uncharitable in others. Afterall, the puzzles and arguments discussed in§2 seem to show that no interpretation can secure the truth ofeverything that we are inclined to say about ordinaryobjects. For instance, the liberal interpretations on which MC1 comesout true (‘Athena and Piece exist’) must, on pain ofcontradiction, make at least one of MC2 through MC4 come out false.But then some other intuitively true claim—perhaps,‘Athena and Piece (if they exist) areidentical’—will come out false. The demandinginterpretations on which MC1 comes out false do better than theliberal interpretations on this score, since they can makeall of MC2 through MC4 come out true. This gain in charitymight then be held to counterbalance the loss in charity fromrendering MC1 false.[61]
One might also challenge CH1 on the grounds that the principle ofcharity, properly understood, demands only that the utterances andbeliefs of ordinary speakers be reasonable, not that they be true.Since itlooks to ordinary speakers as if there are statues,and since they have no reason to believe that appearances aremisleading (having never encountered the arguments for eliminativism),their utterances and beliefs would be reasonable even if false. Theprinciple of charity, so understood, would not favor liberalinterpretations over demanding interpretations.[62]
Another strategy involves resisting the argument at CH2, bymaintaining that there are constraints beyond charity that favor thedemanding interpretations. Charity, after all, is not the only factorinvolved in determining the meanings of our utterances. Certainpuzzles about content determination have been thought to show that thecontent of an expression or utterance cannot be determined solely bywhich sentences we are inclined to regard as true; it is also partly amatter of the relative “naturalness” or“eligibility” of candidate contents. One who accepts thissort of account may maintain that the demanding interpretations,although less charitable, nevertheless assign more natural contents toEnglish sentences than the liberal interpretations, for instance byassigning a more natural meaning to the quantifiers.[63]
Finally, CH3 may be resisted by compatibilists, according to whom whatordinary speakers are saying is compatible with theeliminativist’s claim that statues do not exist. Sinceontological discussions (like this one) are not carried out inordinary English or ordinary contexts, one cannot infer that there arestatues from the fact that ordinary speakers can truly say‘there are statues’, any more than I can infer that I amon the moon from the fact that an astronaut truly utters ‘I amon the moon’.[64]
Arguments from entailment purport to establish that eliminativism isself-defeating, insofar as certain things that eliminativists affirmentail the existence of the very ordinary objects that they wish toeliminate. Here is a representative argument from entailment:
Now consider two arguments for ET2: the argument from identity and theargument from application conditions.
The argument from identity proceeds from the assumption that ordinaryobjects are identical to the smaller objects of which they arecomposed. The statue, for instance, is identical to its atomic parts.Accordingly, by affirming that there are atoms arranged statuewise,eliminativists let into their ontology the very things that they hadintended to exclude.[65]
However, the view that composites are identical to their parts ishighly controversial. One common objection is that the identityrelation simply isn’t the sort of relation that can hold betweena single thing and many things. Another common objection is thatordinary objects have different persistence conditions from theirparts. For instance, the atoms arranged statuewise, unlike the statue,will still exist if the statue disintegrates and the atoms disperse.It would then seem to follow by Leibniz’s Law that the atoms arenot identical to the statue.[66]
The argument from application conditions arises out of some generalconsiderations about how it is that kind terms refer to what they do.Suppose an archaeologist uncovers an unfamiliar artifact, gesturestowards it, and introduces the name ‘woodpick’ for thingsof that kind. Yet there are numerous things before her: the woodpick,the woodpick’s handle, the facing surface of the woodpick, etc.Furthermore, the woodpick itself belongs to numerous kinds:woodpick,tool,artifact, etc. So how is itthat ‘woodpick’ came to denote woodpicks rather thansomething else? (This is an instance of what is known asthequa problem.) It must be because the speaker associatescertain application conditions and perhaps other descriptiveinformation with the term ‘woodpick’, which single outwoodpicks—rather than all tools or just the facing surfaces ofwoodpicks—as the denotation of the term. And the same isplausibly so for already-entrenched kind terms like‘statue’: their reference is largely determined by theapplication conditions that speakers associate with them.[67]
Armed with this account of reference determination, one might thenargue for ET2 as follows. The application conditions that competentspeakers associate with ‘statue’—together with factsabout the distribution of atoms—determine whether it applies tosomething. But these applications conditions are fairly undemanding:nothing further is required for their satisfaction than that there beatoms arranged statuewise. Accordingly, so long as there are atomsarranged statuewise, ‘statue’ does apply to something,from which it trivially follows that there are statues.[68]
This argument may be resisted on the grounds that the applicationconditions that ordinary speakers associate with ‘statue’aren’t quite so undemanding. It’s not enough simply thatthere be atoms arranged statuewise. Rather, there must be an objectthat is composed of the atoms—and (eliminativists might go on toinsist) there are no such objects. However, those who are moved by thequa problem might respond that ‘object’ itself must beassociated with application conditions, which are likewisesufficiently undemanding as to be satisfied so long as there are atomsarranged statuewise.[69]
The material constitution puzzles from§2.3 can be repurposed as arguments against two forms of permissivism:universalism and the doctrine of arbitrary undetached parts (DAUP).The basic idea behind both arguments is that permissivists end upcommitted to objects that are distinct and yet share all of theirparts, which is impossible.[70]
Here is a coincidence argument against universalism. Let the ks be theatoms that presently compose my kitchen table, K, and let us supposethat there is some time, t, long before the table itself was made, atwhich the ks all existed.
CU1 looks to be a consequence of universalism, given our assumptionthat the ks all existed at t. The idea behind CU2 is that there wouldseem to be only two nonarbitrary accounts of the persistenceconditions of the widely scattered fusion F: (i) that F exists for aslong as the ks are in precisely the arrangement that they enjoy at tor (ii) that F exists for as long as the ks exist. Option (i) imposesan implausibly severe constraint on the sorts of changes an object cansurvive, which leaves us with option (ii)—from which it followsthat, since the ks exist now, so does F. The idea behind CU3 is thatthere cannot be distinct objects which (like F and K) have exactly thesame parts and exactly the same location. CU4 is a straightforwardconsequence of Leibniz’s Law: F by hypothesis existed at t, soif F = K then K must also have existed at t. As for CU5, tablesplausibly are essentially tables, in which case K could not haveexisted before the table was made.
Here are some options for resisting the argument (some mirroringresponses to the puzzles of material constitution). One can deny CU2on the grounds thattable is the “dominant kind”,and once K comes into existence, it takes F’s place and F isannihilated. Constitutional pluralists can deny CU3 and affirm that F≠ K, perhaps granting that distinct objects can have exactly thesame parts and location, or insisting that F and K have different(e.g., temporal) parts. Or one can deny CU5, insisting that K is onlycontingently a table and (like F) was once a scattered fusion.[71]
Now for the coincidence argument against DAUP. Take the example ofWoodrow and Woodrow-minus from§2.8. At t1, Molly the cellulose molecule is a part of Woodrowand at t2 Molly is removed and destroyed. LetWoodrow1 be that (if anything) which ‘Woodrow’picks out at t1; Woodrow2, that (if anything)which ‘Woodrow’ picks out at t2; and mutatismutandis for Woodrow-minus1 and Woodrow-minus2.Here is the argument:
CD1 looks trivial: DAUP guarantees that there is an arbitraryundetached part of Woodrow composed of all of its parts other thanMolly. The idea behind CD2 is that Woodrow-minus doesn’t undergoany change between t1 and t2 that could threatenits existence; all that happens is that it is separated from something(Molly) that was not even a part of it. CD3, like CU3, is motivated bythe intuition that objects with the same parts and same location mustbe identical. CD4 is an application of the transitivity of identity:Woodrow-minus1 = Woodrow-minus2, so ifWoodrow-minus2 = Woodrow2, then (bytransitivity) Woodrow-minus1 = Woodrow2; andsince Woodrow2 = Woodrow1, it follows (bytransitivity) that Woodrow-minus1 = Woodrow1.Finally, CD5 is a straightforward consequence of Leibniz’s Law:Woodrow-minus1 and Woodrow1 have different partsand thus cannot be identical.
As with the argument against universalism, one can resist thisargument by denying CD2 and insisting that once Molly is removed,Woodrow-minus is “dominated” by Woodrow and ceases toexist. Or one can deny CD3 and insist that Woodrow and Woodrow-minusare distinct at t2 despite having all the same materialparts at t2. Or one can deny CD4 by insisting that Woodrowhas all of its parts essentially, in which case Woodrow1≠ Woodrow2.[72]
A “gunky” object is a composite object all of whose partsthemselves have parts. The mere possibility of gunky objectsunderwrites an argument against the nihilist thesis that (actually)there are no composite objects.
GK1 is plausible. It seems easy enough to imagine gunky objects, forinstance by imagining an object with a right and left half, each ofwhich itself has a right and left half, which themselves have rightand left halves … “all the way down”, and neverterminating in simple parts. Moreover, it may even be thatactually all objects are gunky. GK2 is trivial: if there aregunky objects in world w, then there is something with parts in w, inwhich case there are composites in w and nihilism is false in w. GK3is plausible as well: the actual world contains what would seem to beparadigm cases of composites (trees, etc.), so if composition occursanywhere, it surely occurs here. Moreover, nihilism is meant to be ananswer to the special composition question, and one would expect suchan answer to be giving necessary and sufficient conditions forcomposition—in which case one would expect proponents ofnihilism to regard it as a necessary truth.[73]
Some will deny GK1. What does seem obviously possible (and easilyimaginable) for there to be certain kinds of infinite descent. Butinfinite descent need not be mereological. For instance, it does seempossible for there to be objects that can be divided into two halves,and whose halves can in turn be divided into two halves, and so on.But it is controversial whether the fact that ocan bedivided into two halves, h1 and h2, entails thato is not simple. One might deny that h1 and h2exist at all before the division: they are brought into existence wheno is divided and, a fortiori, are not parts of o prior to division. Orone might concede that, prior to division, h1 andh2 exist and are partially colocated with o, but deny thatthey are thereby parts of o.[74]
One might instead reject GK3 on the grounds that it runs afoul of“Hume’s Dictum”, according to which there can be nonecessary connections among distinct existences. There is somecontroversy about how best to understand Hume’s Dictum and, inparticular, whether it prohibits necessary connections even betweenoverlapping items (which typically are thought not to be“distinct” in the relevant sense). But assuming that itdoes, then it will rule out any principle of composition according towhich simples in certain arrangements cannot fail to composesomething—for this would be to impose a necessary connectionbetween simples being in that arrangement and the existence of a(numerically distinct) whole that they compose. And if we cannot ingeneral expect theories of composition to be necessary if true, weshould not expect nihilism to be necessary if true.[75]
Just as the possibility of infinite descent can be wielded againstnihilism, the possibility of infinite ascent can be wielded againstuniversalism. Let us say that a world is “junky” iff everyobject in that world is a part of some further object.
The idea behind JK1 is supposed to be that, just as there is nological or conceptual barrier to an infinite descent of parts, thereis no logical or conceptual barrier to an infinite ascent of wholes.(Though not everyone finds themselves able to conceive of junkyworlds.) The idea behind JK2 runs as follows. According touniversalism, every plurality of objects has a fusion, and, inparticular, the plurality consisting of all things has a fusion. Butthere can be no fusion of all things in a junky world. For that fusionwould have to be a part of something (since the world is junky); butif it already has everything as a part, there is nothing left for itto be a part of. JK3 can be motivated in much the same way as GK3:universalism is meant to be an answer to the special compositionquestion, and thus will presumably be necessary if true. Thestrategies considered above for resisting the gunk argument seem toapply equally to the junk argument—one can, for instance, denythat we are imagining what we think we are, or one can invokeHume’s Dictum and deny that universalism is necessary if true.[76]
As we have seen, there are some who deny that ordinary compositeobjects exist, and we have examined some of their reasons forembracing one or another form of eliminativism. But there are alsosome who grant that ordinary objects exist but deny that they existfundamentally. This is an importantly different claim, whichcan be spelled out in either of two importantly different ways.[77]
First, one might deny that any ordinary composite objects arefundamental, that is, one can insist that there is something in whichthey are grounded. Even those who think that ordinary objects existwill likely find it natural to suppose that no ordinary composites arefundamental: all ordinary composites are ultimately going to begrounded in their simple microscopic parts.[78]
On the second understanding of the claim that ordinary objects do notexist fundamentally, the idea is that they are not in the domain of afundamental quantifier, where the fundamental quantifiers are thequantifiers that appear in the best correct and complete theory of theworld. To help see how the two understandings of “existsfundamentally” can come apart, notice that the identity relationis plausibly fundamental (appearing in the best theory of the world),despite the fact that it relates every object—includingnonfundamental objects—to itself. A relation can be fundamentalwithout dragging everything in its extension into the fundamentallevel with it. Similarly, even if the ordinary existential quantifieris a fundamental quantifier, that does not obviously entail thateverything in its domain (namely: everything) is fundamental as well.[79]
Even so, one might deny that the ordinary existential quantifier is afundamental quantifier on grounds of parsimony. Explanations involvingquantifiers whose domains include nonfundamental objects (the ideagoes) will be less parsimonious than explanations involvingquantifiers whose domains include only fundamental objects. And sincethe ordinary existential quantifier includes nonfundamental objects(e.g., ordinary composites), it will be less fundamental thanrestricted quantifiers ranging only over fundamental objects.[80]
How do these stances on fundamentality—that ordinary objects arenot fundamental or are not in the domain of fundamentalquantifiers—compare to the eliminativist theses discussed above?Although there is a superficial resemblance, the differences aremanifest when we consider how the views interact with the argumentsagainst conservatism in§2. Eliminativists, who say that ordinary objects do not exist, canaccept AV5, DK4, OD5, SR4, and ST8 and can reject AR1, MC1, and PM3 inthe arguments above, since the latter affirm, and the former deny, theexistence of ordinary objects. But those who are willing to deny onlythat ordinary objects existfundamentally (in one or theother sense) must find some other way of addressing the arguments.
How to cite this entry. Preview the PDF version of this entry at theFriends of the SEP Society. Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO). Enhanced bibliography for this entryatPhilPapers, with links to its database.
[Please contact the author with suggestions.]
View this site from another server:
The Stanford Encyclopedia of Philosophy iscopyright © 2025 byThe Metaphysics Research Lab, Department of Philosophy, Stanford University
Library of Congress Catalog Data: ISSN 1095-5054