Philosophical Review

Four-Dimensionalism Author(s): Theodore Sider Reviewed work(s): Source: The Philosophical Review, Vol. 106, No. 2 (Apr., 1997), pp. 197-231 Published by: Duke University Press on behalf of Philosophical Review Stable URL: http://www.jstor.org/stable/2998357 . Accessed: 24/08/2012 19:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend access to The Philosophical Review.

http://www.jstor.org

Vol. 106, No. 2 (April 1997) Review, ThePhilosophical

Four-Dimensionalism Theodore Sider

space. A road through timeis like extension Persistence through of the regionof space it occuhas spatialpartsin the subregions partsin an objectthatexistsin timehas temporal pies; likewise, of thetotalregionof timeitoccupies.This subregions thevarious thedoctrine oftemas four-dimensionalism, variously view-known thatobjects"perdure"-is opposedto and the theory poralparts, "endure,"or are thatthings thedoctrine "three-dimensionalism," in favor of thisdispute to resolve "wholly present."'I willattempt based on conbymeansof a novelargument four-dimensionalism in thisarea can be But beforeargument siderations ofvagueness. thanis customI believe becomemuchclearer we must productive, whatthe disputeis, forthe usual waysof forary about exactly wherethree-dimensionare flawed, especially mulating thedispute alismis concerned.
of Predecessorsof parts of this paper were presented at the University of Michigan,and of Arizona, the University the University Massachusetts, the 1994 PacificAPA meetings.For theirhelpfulsuggestionsI would like Carol Cleland, John G. Bennett,Phillip Bricker, to thank Mark Aronszajn, Earl Conee, David Cowles, Fred Feldman, Rich Feldman, Kit Fine, Tove and especialReview, Finnestad,David Lewis,the editorsof the Philosophical ly David Braun and Ned Markosian. (1980), Hughes include Armstrong 'Contemporaryfour-dimensionalists (1986, sec. 5), Heller (1993, 1992, 1990, 1984), Lewis (1986a, 202-4, and 1983, postscriptB), and Quine (1981, 10-13, and 1976). I lump the folcamp, althoughthe view lowingphilosophersinto the three-dimensionalist isn't usuallyclearlyarticulated(all share in the rejectionof temporalparts, though): Chisholm (1976, appendix A), Geach (1972), Haslanger (1994), Mellor (1981, 104), Merricks (1994), Thomson (1983), van Inwagen and longer note 1.11). (1990b), Wiggins (1979; 1980, 25, 25 n. 12, 194ff., objects,such as everyday to identify It is common forfour-dimensionalists planets and persons,withaggregatesof temporal parts-with "space-time worms" as theyare sometimescalled. I disagree: as I argue in my 1996a, temporalparts, objects withthe short-lived everyday it is betterto identify and analyze talk of persistenceover time witha temporalversionof counversionof fourterparttheory.But in thispaper I ignore myidiosyncratic dimensionalismin favorof its more orthodox cousin. 197

THEODORE SIDER

1. What is four-dimensionalism? We need to look carefullyinto just what three-and four-dimensionalism amount to. These names for the doctrines,firstof all, one are poor guides. If by sayingthatan object is four-dimensional means thatit extends throughthe fourthdimension-time-then because nearlyeveryone nearlyeveryone is a four-dimensionalist, agrees that objects persistthrough time. That's not controversial; what's controversial is howtheydo so. Perhaps the sayingthat objects are four-dimensional has a richer meaning, but then we'd doesn't need to knowwhat thatrichermeaning is; the sayingitself suffice.Other obscure phrases are sometimes used to mark the David Wigdistinctionbetween three- and four-dimensionalism. application to anygins,forexample, rejectsfour-dimensionalism's thingother than eventsby sayingthatwhile eventsoccupy periods of time,continuantslike persons don't occupy time,but ratherpertime.2If persisting sist through from throughan intervalis different occupying it, then we need some account of the difference.Yet of the dispute is the claim one someanother poor characterization timeshears thatthe disagreementis over whetheran object at one time is ever "strictly" identical to an object at another.This claim about "strict identity" isn't at all controversial:since everyone identicalwithitself, everyonewho agrees thateveryobject is strictly accepts the basic phenomenon of persistenceover timeaccepts the identical existence of objects that exist at one time and are strictly withobjects thatexistat other times.A finalreason to have a clear statement of the dispute is thatit is sometimessaid thatthe dispute
2Wiggins 1980, 25 n. 12. Wiggins goes on to claim that continuants persistthrough time "gaining and losing parts." But of course the fourdimensionalistwill accept that objects can gain and lose parts-I lose a part x when x's temporal part is part of my temporal part at some time, but x's temporal parts are not parts of my temporal parts at later times. view,thinksof it as a Mark Heller, an opponent of the three-dimensional view according to which a physicalobject is "an enduring spatial hunk of thing,in contrast, times;a four-dimensional matter"thatexistsat different one time untilanother (1990, 4-5). In a similarvein, merelyexistsfrom view,saysthata perPeter van Inwagen, a friendof the three-dimensional during object would have "temporal extent,"whereas the concept of temporal extent does not apply to enduring objects (see 1990b, 252). The these authorsuse are no less obscure than problem is thatthe distinctions theyare attemptthe distinctionbetween three-and four-dimensionalism ing to clarify.
198

FOUR-DIMENSIONALISM

is meaningless,or even merelyverbal! Peter van Inwagen, for example, has said of temporal parts: "I simplydo not understand what these thingsare supposed to be, and I do not think this is myfault.I thinkthat no one understandswhat theyare supposed to be, though of course plenty of philosophers think they do" (1981, 133). And Eli Hirsch has claimed thatthe dispute is merely verbal (1982, 188ff.).3 In thissection I will give a general statementof four-dimensionalism. I hope to phrase that statementin termsthat are clear and acceptable to all concerned so that dispute over its truthwill be neither confused nor meaningless.4 Moreover,the dispute will not be merelyverbal, since the termswill not shifttheir meanings in the mouths of the disputants.To rule out the possibility of obscurity, I'll restrict myself to a meager set of primitive notions. They are, in addition to logical and modal notions,just these two: the mereological notion of a part at a time, and the spatiotemporal notion of existingat a time. Each requires comment. The notion of an object's having a part at a time is familiar:the end of my fingernailis part of me today,but is not part of me tomorrowif I clip it off;a certain plank may be part of the ship of Theseus at one timebut not another,etc. Familiaras thisnotion is, it is not the notion of parthood usuallydiscussed byfour-dimensionalists.FollowingLeonard and Goodman's "Calculus of Individuals" (1940),5 four-dimensionalists tend to speak of the partsof an rather than the parts it has at this time or that. object simpliciter, This is actuallya special case of a more general fact: four-dimensionaliststend to employ an atemporal notion of exemplification of propertiesand relations.Thus, a four-dimensionalist willsaythat inches my currenttemporal part is, atemporally, sitting, sixty-nine will say that this tall, and wearing a hat; and a four-dimensionalist temporal part is, atemporally, part of the larger space-timeworm that is me. This is not to say that four-dimensionalists reject the notion of change. For the four-dimensionalist, change is difference
3I have heard various people claim in conversation that there is no genuine (nonverbal) difference between the views,but Hirsch's is the only claim of thissort in printI know of. 4Haslanger (1994, 340-41) also notes that the obscurity charge can be answeredby mereological definitions, although her statementof the confrommine. troversy differs 5See also Simons 1987, 5-100 on classical mereology.
199

THEODORE SIDER

to standbetween successive temporal parts. I change fromsitting I have a temporal ing, in the intuitivesense of change, because part thatsitsand a laterone thatstands.In a similarsense, I change in whatrelationsI bear: I sitin a chair at one time but not another because myearliertemporalpart sits (simpliciter)in a temporalpart of the chair and one of my later temporal parts fails to sit in the correspondinglater temporalpart of the chair.Similarpointshold for mereological change. My fingernailend ceases to be a part of me , in the intuitivesense of "ceases to be a part of," because its later temporal parts are not part of mylater temporal parts. notions of atemporal We can thinkof the four-dimensionalist's as being those generally, parthood, and atemporal exemplification we take an we employwhen "atemporal perspective"and contemplate the whole of time. But when discussingobjects in time,we We say that the typically do not take this atemporal perspective.6 end of my fingernailis part of me now, despite the fact that I'll then it is not But if I do clip it offtomorrow, clip it offtomorrow. part of me in the atemporal sense, forit has partsthatare not part its futuretemporal parts afterthe clipping). The of me namelyy, ratherthan atemporal: everyday notion of parthood is temporary, is of me now. This is not to say thatthere's the fingernailend part use of the atemporal wrongwiththe four-dimensionalist's anything notion of parthood. A four-dimensionalist can take the atemporal notion as basic, and characterize temporaryparthood using that
notion :7

x and y each existat t,and x is partof y at t iff (P@T) Necessarily, x's temporalpart at t is part of y's temporalpart at t. parthood the simplytreatstemporary Here the four-dimensionalist same way that she or he generallytreatstemporarypropertyexmay admit that sometimeswe use 'part of 6Even three-dimensionalists when talkingabout thingsthatare not in time in the in an atemporalway, same waythatcontinuantsare. Examples mightinclude talkof timesthemcentury"),events,or allegedly selves ("the 1960s are part of the twentieth atemporal things ("arithmeticis part of mathematics"). The relationship between these uses of 'part' and the notion of parthood I utilize in the textis controversial. can replyto Ali Kazmi's complaint that 7Thus, the four-dimensionalist implies that,for example, my fingernailend is not four-dimensionalism part of me-the replyis that my fingernailend is part of me now,in the sense described in the text.See Kazmi 1990, 231 n. 3. 200

FOUR-DIMENSIONALISM

amplification.Sittingat a time, recall, is simplytaken to involve temporalpart which is located at thattime;having havinga sitting is having a temporal part located at thattime a time at a part x as which contains x's temporal part then as a part. parthood characterizestemporary Thus, the four-dimensionalist in conin termsof atemporal parthood. The three-dimensionalist, temporal to since it appeals this characterization, will reject trast, parthood is irreducibly temporary parts;fora three-dimensionalist, notion of atemrelative.As forthe four-dimensionalist's temporally poral parthood, at least as applied to objects that persistthrough is likelyto deny that it has sense. time, the three-dimensionalist One cannot say that my arms and legs are part of me simpliciter; relation one must alwaysspecifythe time at which the part-whole obtains. This is an instance of a more general pattern:while the propertiessuch as being will say that temporary four-dimensionalist fivefeettall are had simpliciter(by temporal parts), the three-dimensionalistwill insistthatsuch "properties" are had onlyrelative to times.8Of course, everydayuses of 'part' could be missing a means 'part but in such cases 'part of' implicitly temporalqualifier, of now'. This differencein how the notion of parthood is understood raises a problem for typical statementsof four-dimensionalism. Those statementsare phrased in terms of atemporal parthood, theyare incomlights, whichmeans thatbya three-dimensionalist's
intrin8This factformsthe basis forLewis's argumentfrom"temporary must accept a sics" (1986a, 202-4). My claim that three-dimensionalists I am temporallyqualified notion of parthood is actually oversimplified; in This is manifested ignoringthe viewof those who "take tense seriously." my assumptionthat all propositionshave permanent truthvalues; I've assumed thatifparthood is not atemporal,then the notion of parthood must be the notion of havinga part at a time. But one who takes tense seriously would have a thirdoption: even though parthood is not atemporal, 'x is proposition,which in some part of y' expresses a complete present-tense One sense is not reducible to eternal propositions about parthood-at-t. holds that there are no objects particularversion of thisview,presentism, exist; for ease of exposition I'm ignoringpresentism that don't currently as well. I ignore these viewsbecause (contra Merricks1995) I take them see my 1996b. My to be independent of the truthof four-dimensionalism; withirreducargumentswould simplyneed to be restatedin a framework ible tense. way of thinkingabout atemKit Fine (1994) discusses a quite different poral parthood (and other topics related to the present paper), which I will not discuss in thispaper. 201

THEODORE SIDER

plete, in somethinglike the way 'John is ten feet from' is incomintelligibleif four-dimensionplete. Of course, theyare perfectly alism is true. But it is desirable to stateopposing viewsin a neutral of the language, so that the opponents may agree on the identity propositionunder dispute. Moreover,we do not want to hide fourdimensionalismin the verylanguage we use to raise the question of its truth.So I suggestwe employ a language in which mereological concepts are temporallyqualified; let us say 'part of at t' will disagree instead of 'part of. Three- and four-dimensionalists over whethertemporaryparthood can be accounted for in terms of atemporal parthood (via (P@T)), but will agree on the intelliparthood; thus,the framework of the notion of temporary gibility is neutral. of temporaryparthood relative,then so must be certainother If parthood is temporally mereological notions that can be defined in terms of parthood. speak of objects overlapping (sharing a part Four-dimensionalists in common), and of the mereological fusion,or sum, of a class of objects (a fusionof class S is an object thatcontains everymember of S as a part, and is such that each of its parts overlaps some we must speak of objects member of S); in our neutral framework overlapping at a given time, and of an object being a fusion of a class at a specified time.9
9For tensed mereology,the basic notion can be taken to be parthoodsomethingis part of each then.Where Two objects overlap at a timeiff at-t. S is a class of objects that exist at t, x is a fusion of S at t iff (i) every some member of S is part of x at t, and (ii) everypart of x at t overlaps-at-t is transitive, thateverything member of S. I will assume thatparthood-at-t that existsat t is a part of itselfthen, that x is part of y at t only if x and principle is true: y both exist at t, and that the following (PO) If x and y exist at t, but x is not part of y at t, then x has some part at t that does not overlap y at t. (This is the temporal analog of a theorem of the Calculus of Individuals; see SCT13 fromSimons 1987, 38.) See also Thomson 1983, 213-220,and Simons 1987, 175ff. principle," The atemporalcalculus of individualscontained the "identity but I do notassume according to which mutual parthood entails identity; its temporal analog, that no two objects can be parts of each other at a givtime. This principlewill clearlybe rejected by the four-dimensionalist, en that 'part-at-t' obeys (P@T), for any two space-timewormsthat share a temporalpart at some time provide a counterexample to each. And even coinciding statuesand lumps sometimesdistinguish three-dimensionalists then the statueand the lump would be partsof each of clay;if thisis right, other.At the time of coincidence, any part of eitherwill share subatomic 202

FOUR-DIMENSIONALISM

In order to state four-dimensionalism I will need, in addition to parthood-at-t, a temporal notion of existence-at. The notion is familiar: I exist at the present time but not at times before 1967; Socrates existed in the distantpast but does not existat the present time,etc. As withtemporary parthood, there is a principlegoverning existence-at thatis accepted by the four-dimensionalist but not an object exists at a time iffit has a by the three-dimensionalist: temporal part that existsat that time.10 I distinguish existence-at fromquantification. When I say simply "there is," I intend atemporal quantificationover all objects, not is analjust those that are located at any particulartime. Exists-at ogous to the spatial predicate 'is located at', rather than to the logician's ']'. There is a view in the philosophyof time which opposes this notion of atemporal quantification.I say that there is such a thing as Socrates, which doesn't exist at the currenttime;
particlesin common with the other,but then, by (PO), theyare parts of each other at the time. mightle101nfact,there is a distinctnotion that a four-dimensionalist gitimately call "existence at." (For this point, see also Heller 1984, 32829.) In this other sense, I do not exist at the present time because I do not whollyexistat the presenttime-that is, because I have parts (namely, futuretemporalparts) thatdo not exist at the presenttime. On the sense of 'existence at' in the text,an object gets credit for existingat a certain time in virtueof its having a mere temporal part that existsat that time. fromHeller's-he uses one notion My notion of existence-atthus differs where I use the other. I choose my usage so that three-and four-dimenIt is the sense in the sionalistscan accept the same notion of existence-at. sense of existence text,not Heller's sense, thatcorrespondsto the everyday at a time,for on Heller's sense I do not exist at the present time (since I exist). Thus, the four-dihave futuretemporal parts that don't currently mensionalist will accept that the term 'existsat', as used in the text,obeys the law: (E) Necessarily, an object x existsat time t iffsome part of x existsat t the truth (Note that'part' here is atemporal.Thus, (E) does not contradict as follows:"the parts of an object sometimes that may be put intuitively notion of partoutlivethatobject," for that truthconcerns the temporary hood.) Additionally, in this paper 'some object' and 'every object' range only over thingsthatexist in time; thus,I assume 'existence-at'to be governed by the following principle: (T) Necessarily, each object existsat some time I'll also assume the existence of a set of all the thingsthatexistin Finally,
time.

203

THEODORE SIDER

there simplyis no such object as Socbut according to presentists, existingobjects. In thispaper presently are rates. The only objects I'll assume that presentismis false, but this is only to avoid complication: the claims here could all be restatedwithina presentist 1I1 framework. Given just these two notions, existence-atand temporarypartAs hood, we can give a general statementof four-dimensionalism. is the claim thatthe partI see it, the heart of four-dimensionalism whole relation behaves withrespect to time analogously to how it spatial behaves withrespect to space: just as thingshave arbitrary temporalparts.When applied to parts,theylikewisehave arbitrary that partsmeans, roughly, space, the idea thatthingshave arbitrary for any way of dividing the region of space occupied by a given object, there is a correspondingwayto divide thatobject into parts that exactlyoccupy those regions of space.12 Applied to time, the of an object into idea is thatforanywayof dividingup the lifetime separate intervalsof time,there is a correspondingwayof dividing the object into temporal parts that are confined to those intervals of time. This latter claim may be captured in a concise thesis as span of an object, x, is the set of timesat follows.Say that the time whichx exists;and suppose thatwe have twodisjointsubsetsof the time span of x, T1 and T2, whose union is T. What we want to say is that there are two objects, xl and x2, whose time spans are T, that "add up to" x. But since we are using and T2, respectively, we cannot simplysay that qualified mereological terms, temporally of xl and x2. Instead we say that at x is the fusion, atemporally, everymoment of T1, x and xl have the same parts; and similarly for T2 and x2. We have, then, the following forany object x, and for Necessarily, Locality: Thesisof Temporal sets of timesT1 and T2 whose non-overlapping any non-empty, union is the time span of x, there are two objects xl and x2, such that (i) xl and x have the same parts at everytime in T1, (ii) x2 and x have the same parts at everytime in T2, and (iii) the time span of xl = T1, while the time span of x2 = T2.13
and my 1996b on the "ISee Prior 1968 and Adams 1986 on presentism, independence of presentismand four-dimensionalism. '2Some reject this thesis.See van Inwagen 1990a, 74-80, for example. 13Here and throughoutthis paper I ignore the view that ordinaryobjects contain immanentuniversalsas parts.My account is similarto thatof Heller 1984, 325-29. See also Lewis 1983, postscriptB; Thomson 1983,
204

FOUR-DIMENSIONALISM

captures Evidence thatthe Thesis of Temporal Localitycorrectly want to say comes fromthe factthatthe what four-dimensionalists Thesis of Temporal Localityentails the doctrine that objects have need a definitionof 'temporal part'. The temporal parts.We first temporal part of x at time t is sometimesdefined as the part of x that exists only at t and has the same spatial location as x; but I distrustthe appeal to spatial location. The idea is to ensure that the temporalpart of x is a "big enough" part of x; but the spatial definition failsfor objects withoutspatial location. The spatial definitionwould also fail if an object had multipleparts thathad the same spatial location as it (if an object had as a part a "trope" corresponding to its shape, this should not turn out to be a temprefera purelymereological poral part of thatobject). I therefore definition:
part of y at instantt = df (i) x exists x is an instantaneous temporal x of y at t, and (iii) x overlaps at t is at part t, (ii) but only at, that is part of y at t. everything

This captures the idea that mycurrenttemporal part should be a part of me now that exists only now, but is as big as I am now. It should overlap my arms, legs-everything that is part of me now. Though this characterizesinstantaneoustemporal parts,we could generalize to consider extended temporal parts:an extended temporal part of x throughoutintervalT is an object whose time span is T, which is part of x at everytime during T, and which at every thatis part of x at thatmoment. moment in T overlaps everything by 'temporal part' I'll mean 'inUnless otherwisenoted, however, of 'temporalpart', stantaneoustemporalpart'. Given mydefinition
206-10; and van Inwagen 1981, 133 and 1990b, 245-48. The version of strongone, since it imI have stated is a particularly four-dimensionalism subset of the time set of an plies the existence of a temporalpart for every object. It allows,for example, instantaneoustemporalparts and temporal partswithradicallydiscontinuoustemporallocations. One mightargue for of various kinds, for example to temporal divisions that are restrictions continuous intervals.See Wig"natural" in some sense, or to temporally vergins 1980, 24-27 and Mellor 1981, 132-34. I preferthe unrestricted withwhichI have no real quarsion, although thereare certainrestrictions rel (see section 3.3). Moreover,the prefix'necessarily'will strikesome as possibility too strong.David Lewis, for example, accepts the metaphysical times (see the introof an object's being whollypresent at two different a restricted duction to Lewis 1986b, x). Haslanger (1994, 340) formulates versionof four-dimensionalism. 205

THEODORE SIDER

that the Thesis of Temporal Locality has the desired entailment14 an object must have a temporal part at everymoment that it exwormscorrespondists.15(Thus, someone who accepted space-time temporal from,"whollypresentobjects" without ing to, but distinct on myusage.)'6 partswould not count as a four-dimensionalist, could offeratemporalanalogs of Note thata four-dimensionalist of 'temporalpart':'7 the Thesis of Temporal Localityand definition non-overfor any object, x, and forany non-empty Necessarily, time span is the union T2 whose lapping sets of times T, and of x, there are two objects x, and x2, such that (i) x is the fusion of x, and x2,and (ii) the time span of x, = T1, whereas the time span of x2 = T2
part of y at instantt =df (i) x is a x is an instantaneous temporal part of y, (ii) x existsat, but onlyat t, and (iii) x overlapsevery part of y that existsat t

Relative to this atemporal definitionof 'temporal part' (and the assumptionsabout atemporal parthood fromLeonard and Goodman's Calculus of Individuals), the atemporal Thesis of Temporal Localityentails thateveryobject musthave a temporalpartat every moment that it exists;'8it also has the consequence thateveryob14Here and elsewhere,when I say that somethingis "entailed by" or "follows from" something else, I mean that the formerfollowslogically the assumptionsabout mereologyand fromthe conjunction of the latter, existence-atthat I make explicit in notes 9 and 10, and the principlesof standardset theory(including the axiom of choice). 15Proof: Consider any object x at any moment t of its career. If x exists only at t, then x is obviouslya temporal part of x at t. Otherwise,apply the Thesis of Temporal Localityto the sets {t} and the set consistingof all the members of the time span of x except for t-the resultingxi is our x and xl have the desired temporal part. For xl existsonly at t; moreover, xl is a part of x at t is reflexive, same parts at t. Thus, since parthood-at-t that is part of x at t. thatoverlaps at t everything 161 thankan anonymousrefereefordrawingmyattentionto thispoint. 17A more general versionof the atemporal Thesis of Temporal Locality of x's timespan, there would assertthatfor any(possiblyinfinite)partition existsa (possiblyinfinite)set of objects,of which x is the fusion,the members of which are confinedto the correspondingmembersof the partition. the more general In the case of the original Thesis of Temporal locality, followsfromthe simplerone stated in the text (withthe help formulation of the axiom of choice), which in turnfollowsfroma stillsimplerversion claiming that for any single subset of x's time span, there is an object confined to that subset that shares parts with x throughout.I thank Kit Fine for these observations. a temporal '8Suppose thatx existsat t. If x existsonly at t, x is trivially 206

FOUR-DIMENSIONALISM

ject is the fusion of its temporal parts.'9In what follows,however, I'll thinkof four-dimensionalism as being stated using temporally qualified mereological terms. Notice thataccording to mydefinition of 'temporal part', a temporal part of x at t must literally be part of x at t. Temporal parts so defined must thereforebe distinguishedfromwhat we might call "ersatz temporalparts,"pairs of objects and timesforinstance. While (x,t) may be suitable to play part of the role that the temporal part of x at t is supposed to play,20 many philosophical uses of temporal parts require that temporal parts literally be parts of objects. This is particularly clear in the use of temporal parts in solving the traditionalparadoxes of co-located objects. For example, it is said to be possible for a statue and the lump of clayfrom which it is made to share spatial location because theyoverlap by sharing temporal parts. But if the temporal part of x at t were simply(x,t), then numericallydistinctobjects could never share a single temporalpart,forwheneverx and y are distinct, so are (x,t) and (yt). should be contrastedwith other docMy four-dimensionalism trines that sometimes go by the same name. Some may use the termfor the view that time is a "fourthdimension," analogous in various waysto the spatial dimensions; my usage is narrower, and concernsjust one analogybetween timeand space concerningpersistence and parthood. Moreover,on my usage, four-dimensionalpart of itselfat t. Otherwise,apply the Thesis of Temporal Localityto the intervalsjt} and the time span of x-jt}; the resultingxl is our desired temporalpart. For xl clearlyexistsonlyat t. Moreover,since x is the fusion of xl and the resultingx2, xl is part of x. Finally, let w be any part of x thatexistsat t; we mustshow thatxl and w overlap.w musthave a partw' thatexistsonly at t (w' is w itself ifw existsat no timeotherthan t; otherwise apply the Thesis of Temporal Localityto obtain w'.) Since w' is a part of x and x is the fusionof xl and x2,w' mustoverlap eitherxl or x2 at t. But wI cannot overlap x2,forby (E) and (T) (note 10), x2would then exist at t. So w' overlaps xi, and hence w overlaps xl. 19Clearly, each of x's temporal parts is a part of x. It remains to show thateverypart of x overlaps some temporalpart of x. Let y be any part of x. By (T) (note 10), y existsat some time t; by (E) (note 10) x existsat t as well, and so (see previousfootnote) has a temporalpart z then; by the (atemporal) definitionof 'temporal part', z overlapsy. 20Ersatztemporal parts are perhaps all we need for the task of stating the search for criteriaof identity over time as the search forunityor genidentity relationsbetween temporalparts. (See, forexample, the introduction to Perry1975.) 207

THEODORE SIDER

ism does not implythat factsabout temporal parts are "prior to," or more "fundamental" than, factsabout continuants.It does not imply that continuant objects are in any sense constructedfrom over time is their temporal parts. Nor does it implythat identity "reducible to temporal parts," in the sense of David Lewis's "Humean Supervenience." Humean Supervenience implies that all facts (in worlds suitablylike the actual world) supervene on the distributionof "local qualities" throughoutspacetime; but local qualities would be instantiatedby temporal parts; and so facts over about temporalpartswould determineall factsabout identity time. The Thesis of Temporal Locality implies no such supervenience; it merelyimplies that the temporal parts must exist. In particular,the Thesis of Temporal Locality is consistentwith there and so the relations, being nonqualitative"unity,"or "genidentity" rotatinghomogeneous disk/sphereis no counKripke/Armstrong etc. are imreducibility, These questions of priority, terexample.21 portantquestions about temporalparts,but theymustbe separated from the more basic question of whether temporal parts exist at that is my concern. all. It is thus "minimal" four-dimensionalism 2. What is three-dimensionalism? whose deI turn now to the statementof three-dimensionalism, fenders deny the analogy between persistence through time and spatial extent. Friends and foes alike often characterizethat doctrineas the viewthatan object is "whollypresent" at everymoment of its existence. We have D.H. Mellor: lifetimes. their (1981,104) are wholly throughout present things Peter Simons: a continuant is wholly (1987, present. At anytimeat whichit exists,
175)

George Graham, who though he rejects three-dimensionalism thinksit our "everyday"view: exists thewhole a person atwhich at anytime ... that we usually think at thattime.(1977,309) or entire personexists
to Lewis 1986b on Humean supervenience.Krip21See the introduction sphere is diske's example is from an unpublished lecture; Armstrong's cussed in his 1980, 76-78.
208

FOUR-DIMENSIONALISM Paolo Dau: conception, the entireobject is to be found On the three-dimensional at each instantthat it exists. (1986, 464) David Wiggins: questions of continuityand persistence that perplex our habitual modes of thought about identityand difference... [need] answers given in language that speaks as simplyand directlyas natural lancontinuants-things with guages speak of proper three-dimensional spatial parts and no temporal parts,which are conceptualized in our whole experience as occupyingspace but not time, and as persisting time. (1980, 25) through And finally,David Lewis: iff,somehow or other, it exists at Let us say that something persists iffit various times; this is the neutral word. ... [Something] endures persistsby being whollypresent at more than one time. ... [M] any would favorthe viewthat [a person, e.g. Hume] endures,whollypresent at everytime of his life,so that those times ... overlap by having him as a shared part. (1986a, 202 and 210)22 The suggestion in these quotations, then, seems to be that for any (continflant) x, and any time t, if x exists at t then x is "wholly like perpresent" at t. (The restriction to "continuants"-things necessary etc.-is sons, planets, protons, medium-sized dry goods, since three-dimensionalists often admit that someobjects, events for example, do have temporal parts.) This is an unfortunate way to formulate three-dimensionalism. What is it for an object x to be "wholly present" at time t? The idea is presumably that every part of x must exist at t. But every part at what time? For three-dimensionalists, the parthood relation is temporally relative, and so "every part of x exists at t" is incomplete since 'part of is temporally unqualified. There are various ways to fill in the temporal qualifier. We might 22Simons'spassage comes rightafterwhat seems to be a definition:"a continuantis an object which is in time,but of which it makes no sense objects, such as taever,that Simons means to be assertingthat everyday bles, chairs,people, etc., are continuants. the temporal qualifier; Lewis has not made the mistake of forgetting and withinhis own framework, rather,he is statingthree-dimensionalism relation presupposition:that the part-whole with an unfriendly therefore
is atemporal. to say that it has temporal parts or phases." It is clear from context, how-

209

THEODORE SIDER

take the claim thatx is whollypresentat t to mean thateverything that is part of x at t exists at t; the slogan then amounts to the following: foreveryx and everytime t at whichx exists, (WP1) Necessarily, everypart of x at t existsat t. trivial-no one would The trouble with (WP1) is that it is utterly deny that a part of an object at a given time must exist thenwhereas the claim thatan object is whollypresentat everymoment between of its existence is supposed to marka point of controversy who defined A four-dimensionalist three-and four-dimensionalists. parthood via (P@T) would accept (WP1). temporary A more likelysense of 'whollypresent',myintended sense from now on, may be defined as follows: (WP) that is at any time x is wholly at t =df everything present part of x existsand is part of x at t.

whollypresent But on thisreading the claim thatobjects are always becomes, foreveryx and everytime t at whichx exists, (WP2) Necessarily, thatis a part of x at some time or other exists everything and is part of x at t. for (WP2) entailsthe too strong, This makes three-dimensionalism of gain and loss of parts. Granted, somethree-dimenimpossibility sionalistswould accept this consequence, most notably Roderick Chisholm (see his 1976, app. B). But most three-dimensionalists and so mereological essentialism are not mereological essentialists, shouldn't be builtintothe statementof three-dimensionalism. Here are some theses (note What,then,is three-dimensionalism? that 'whollypresent' in (WP6) and (WP7) is intended in the sense of (WP)):23 there are no temporal parts. (WP3) Necessarily, nothing that existsfor more than an instant (WP4) Necessarily, ever has a temporal part. nonforany object x, and foranynon-empty, (WP5) Necessarily,
that I consider (WP5). 23I thank an anonymousrefereefor suggesting

210

FOUR-DIMENSIONALISM

overlapping sets of times T, and T2 whose union is the time span of x, there are no two objects xl and x2, such that (i) xl and x have the same parts at everytime in T1, (ii) x2 and x have the same partsat everytimein T2, and (iii) the time span of xl = T1, while the time span of x2
-

T2.

(WP6) In the actual world, small particles (for example, electrons) are whollypresent throughouttheirlifetimes. (WP7) It is possible that some object is whollypresent at more than one time. None is completely satisfactory as a general statementof threedimensionalism. (WP3) cannot be a correct statementof three-dimensionalism of inwill admit the possibility because many three-dimensionalists stantaneous objects, objects that appear only for an instantand then disappear. Such objects would be temporal parts of themselves. As for (WP4), imagine a lump of clay that gets made into a statue-shape foronlyan instant(by a god, say). It seems to me that mightwant to say that in that instant, some three-dimensionalists a statue comes into being, but immediately goes out of existence. Afterall, many three-dimensionalists say thatwhen a lump of clay becomes statue-shaped forsome extended period of timeand then gets squashed, a statue comes into being for that period of time; the instantaneousstatue would be a limitingcase. I'm not myself claimingthatinstantaneousstatuesare possible, but it seems to me that theyaren't inconsistentwith the "picture" three-dimensionalists seem to accept, and so shouldn't be ruled out automatically But this case would viby our statementof three-dimensionalism. olate (WP4), for the statuewould be a temporal part of the lump. that As defined above, a temporalpart of the lump at t is anything (i) is part of the lump at t, (ii) exists only at t, and (iii) overlaps thatis part of the lump at t. Condition (ii) is clearly at t everything satisfied.As for condition (iii), at the time in question, the lump and the statue are made up of the same subatomic particles;thus, thatis part of the lump thenwillshare subatomicparticles anything withthe statue. Finally, condition (i) can be argued foras follows. 211

THEODORE SIDER

relativizedanalog of a principlefrom is a temporally The following the Calculus of Individuals (see note 9), and is surelycorrect: (PO) If x and y exist at t, but x is not part of y at t, then x has some part at t that does not overlap y at t. As I just mentioned, the statue and the lump at the time in question are made up of the same subatomic particles;thus,everypart of the statue at t will,at t, share subatomic particleswith,and thus
overlap, the lump. By (PO), (i) then follows.24

A similarexample shows (WP5) to be unsuitable. Suppose a cerin tain lump of clay is created in statue shape, and afterpersisting stataltered to a distinct thisformfora while,gets instantaneously ue shape, which it retainsuntil being annihilated some time later. willwant to say thatin thisexample, in Many three-dimensionalists addition to the lump of clay we also have two statues, one that comes into being when the lump is created, and another that restatue at the time the lump changes shape. If so, places the first then we have a violation of (WP5), forargumentssimilarto those given in the previous paragraph establish that the lump has the portion same parts as the firststatue at all times during the first of its life,and has the same parts as the second statue at all times during the second portion. (WP6) is a more likelycandidate, but I stillhave mydoubts. First, its restriction to small objects makes it too weak to count as a genof three-dimensionalism? as a formulation 24Whatof the following nothingthatexistsformore than an instanteverhas (WP4') Necessarily, a temporal part at everymoment of its existence. I would reject this statementalong with (WP4), because of a modified version of the example in the text. First,if it is possible for time to be discrete,then we could imagine a lump of clay that takes on a radically different statuesque shape at each momentof its existence.In such a case, a three-dimensionalist mightwant to hold that a distinctstatue is constituted by the lump of clay at each moment of the lump's existence; but these statueswould be temporalparts of the lump, falsifying (WP4'). And even if it is impossible thattime be discrete,a more exotic example might still be possible, in which a three-dimensionalist mightwant to say that (WP4') is violated. Imagine a certain lump of claywitha radicallydiscontinuous shape throughoutits entirecareer.At everyinstantt of its life,(i) the lump has some statuesque shape S at t, and (ii) there is an intervalof if the lump exists time about t, such that at everymoment in the interval, at that moment,the lump has at thatmoment a shape thatis quite different fromS. 212

FOUR-DIMENSIONALISM

selThree-dimensionalists eral statementof three-dimensionalism. dom confine their remarks to subatomic particles; they say that macroscopic objects such as persons are whollypresent over time. disbold. What if scientists Secondly, (WP6) seems too empirically in flux,exchanging covered thatsubatomicparticlesare constantly threepartsat everymoment?Would those who accept the intuitive dimensionalistpicture need to change their minds? The impresvan Inwagen,etc. is thatthreesion one getsfromreading Wiggins, by such empirical research. dimensionalismwould not be falsified Moreover,no such thesis about actualitywould be a conceptual over time.25 thesisabout the nature of identity The final and weakest thesis on the list, (WP7), will, I believe, for three-dimensionalists be accepted by all three-dimensionalists, willaccept thatwhile persons infactgain and lose parts,theymight not have; and while it could be that subatomic particles are conin mereological flux,it is at least possible thattheyare not. stantly But there is a nagging feeling that something is missing. (WP7) applicable, positiveclaim about the does not contain a universally essential nature of identityover time! Is the positive picture of identityover time one gets from reading the writingsof threea mere mirage? dimensionalists A three-dimensionalist might give up on the attemptto give a mereological account of an object's being whollypresent and understand that notion in some other way. One wonders whether 'whollypresent' would then be an apt term. Moreover,attempts belike this tend towardsthe obscure: recall Wiggins'sdistinction tween occupying a region of time and persistingthrough that region (and see note 2). But regardlessof non-mereologicaldisputes we could consider,we do have a clearlyformulatedmereological dispute at hand that is worth considering: that of whether fourdimensionalismis true. For whateverelse theythink,three-dimensionalistsreject the Thesis of Temporal Locality,togetherwithits implicationthat all objects must have temporal parts. 3. In defense of four-dimensionalism in is often supported by appeal to its utility Four-dimensionalism over time. While solvingvarious traditionalpuzzles about identity formulates whatshe calls the "endurancetheory"as a 25Haslanger of themeaning claimaboutactuality (1994,340), butshe does notclarify wholly present'. 213

THEODORE SIDER

I would like to develop a I fullyendorse this line of reasoning,26 new,more directargument. mereological composition 3.1 Unrestricted will be parallel to an arguMy argumentfor four-dimensionalism mereologment fora different thesis:the principle of unrestricted ical composition, according to which any class of objects whatsoever has a fusion. In the present section I develop in detail the composition,and then in the final two argumentfor unrestricted I show how a sections parallel argumentfor four-dimensionalism most maybe constructed.The argumentsmake some assumptions, notably that vagueness never results from "logic" (that is, from or identity).Though I do not boolean connectives,quantification, say that my assumptionscannot be coherentlydenied, I do think the assumptionsare plausible; I also suspect that theyare widely held, even among those hostile to temporal parts.There is, therein showingthatanyone who accepts the fore,considerable interest assumptionsmust accept four-dimensionalism. composition A starting point for the argumentfor unrestricted that I will develop is David Lewis's argumentfor the same conclusion:

sumsof things thatcontrast We are happyenoughwith mereological and that morethanthey do with one another; their with surroundings to stick and actjointly. We are morereluctant are adjacent, together, are disparate sumsof things that affirm of mereological theexistence ... and go their and scattered separate ways. is as follows.... To restrict restricted The trouble with composition a vague our intuitions wouldrequire in accordance with composition thenit must restriction. Butifcomposition obeysa vaguerestriction, takes sometimes be a vaguematter whether place or not. composition And thatis impossible. locatesit in our thought The onlyintelligible accountofvagueness The reasonit'svaguewheretheoutback and language. beginsis not rather withimprecise thatthere'sthisthing, the outback, borders; and nobodyhas been withdifferent thereare many borders, things, a choiceof one of themas theofficial foolenoughto try to enforce But is semantic indecision. of theword'outback'. Vagueness referent connectives aren't, not all of languageis vague.The truth-functional and difference, and for forinstance. Nor are the wordsforidentity
26I

in my1996a. fortemporal givethissortof support parts 214

FOUR-DIMENSIONALISM the partial identity of overlap. Nor are the idioms of quantification, so long as theyare unrestricted. How could any of these be vague? What would be the alternatives between which we haven't chosen? The question whether composition takes place in a given case, whethera given class does or does not have a mereological sum, can be stated in a part of language where nothing is vague. Therefore it cannot have a vague answer.... No restriction on compositioncan be vague. But unless it is vague, it cannot fitthe intuitivedesiderata. So on composition can serve the intuitionsthat motivate no restriction it. So restriction would be gratuitous.Composition is unrestricted.... (1986a, 212-13) LO-Wis's version of the argument may be summarized as follows. (I follow Lewis in speaking of parthood atemporally; I consider temporally relativized parthood in the next section.) If not every class has a fusion, then there must be a restriction on composition. Moreover, the only plausible restrictions on composition would be vague ones. But there can be no vague restrictions on composition, because that would mean that whether composition occurs is sometimes vague. Therefore, every class has a fusion. My version of the argument will take a differentform. My reason for abandoning Lewis's version of the argument is its apparent assumption, reproduced as the firstpremise of my summary, that if not every class has a fusion then there must exist a "restriction on composition." On a natural reading, a "restriction on composition" is a way of filling in the blank in the following schema:
A class, S, has a fusion if and only if

such that what goes into the blank is not universally satisfied. That is, a restriction on composition would be an answer to Peter van Inwagen's "special composition question."27 (For example, one answer might be that a class has a fusion iff its members are "in contact".) But thus understood, the firstpremise is subject to the following objection: Perhaps the special composition question has no informative answer because whether composition takes place in a given case is a "brute fact" that does not admit of informative
analysis.28

27Actually, the Special Composition Question is slightly since different, it concernswhen fusiontakesplace at a giventime;see van Inwagen 1990a, chapter 2. 2cThanks to David Cowles and Ned Markosianhere. Markosiandefends thisclaim about composition in his "Brutal Composition" (forthcoming).
215

THEODORE SIDER

There are two senses in which composition mightbe brute,one strongerthan the other.In the strongsense, compositiondoes not even supervene on causal and qualitativefactors.This seems extremely implausible. How could there be two cases thatare exactly alike in termsof causal integration, qualitativehomogeneity, etc., but such thatobjects have a sum in one case, but not in the other? But even ifsupervenience is admitted,compositionmightbe brute in the weaker sense that there is no natural, finite,humanlystaon composition. Since I do not wish simplyto teable restriction reject weak brute composition out of hand, I will approach the argumentin a different way. Let us understand a "case of composition," or simplya "case" for short,as a possible situationinvolvinga class of objects having certain properties and standing in certain relations. We will ask with respect to various cases whethercomposition occurs, that is, whetherthe class in the case would have a fusion.In summary, my as If class has a then we argumentruns follows. not every fusion, can consider two possible cases, one in which composition occurs and another in which it does not, which are connected by a "continuous series of cases," each extremelysimilar to the last. Since composition can never be vague, there must be a sharp cutoffin thisseries of cases where compositionabruptly stopsoccurring.But that is implausible. So composition alwaysoccurs. Let us develop the argumentmore carefully, beginningwiththe idea of a continuous series of cases. Firstconsider any case, C1, of which many would say that composition occurs in it-the case of a certain class of subatomic particlesthat are part of mybody,for example. Now consider a second case, C2,which occurs afterI die and am cremated,in which mymolecules are scatteredacross the MilkyWay. Some would say that in C2, composition fails to take place: there is nothing that is made up of these scattered,causally unconnected particles.Next, let us further imagine a finiteseries of cases connecting C1 and C2, in which each case in the series is similarto its immediately extremely adjacent cases in the series in all respectsthatmightbe relevantto the question of whethercomof position occurs: qualitativehomogeneity, spatial proximity, unity action, comprehensivenessof causal relations, etc. I call such a series a "continuous series connecting cases C1 and C2."
216

FOUR-DIMENSIONALISM

premise can now be stated as follows: My argument'sfirst P1: If not everyclass has a fusion,then there must be a pair of cases connectedby a continuousseriessuch thatin one, compositionoccurs,but in the other,compositiondoes not occur. is based on the claim I can thinkof only two objections. The first occurs; for if there are never any cases of that composition never composition at all, then there will be no continuous series conOn this view,which necting a case of composition to anything.29 Petervan Inwagen calls "nihilism,"thereare no compositeobjects. Peter Unger defends a near relativeof thisview,and van Inwagen It deserves spedefends the view in the case of nonlivingthings.30 cial mention because it admits of a betterdefense than one might think.Van Inwagen points out thatthe shockingconsequence that speaking exist does not preclude tables and chairs do not strictly ordinaryassertionsabout tables and chairs being at least loosely speaking true, since theyare paraphrasable as complicated assertions about the fundamentalparticlesthat "compose" the "tables forit depends and chairs." However,thisresponse is unsuccessful, for its success on the a prioriassumption that the "objects" of our ontology (tables, chairs,etc.) are composed of mereologeveryday ical atoms-things withoutproper parts. This assumptionneedn't is that electrons,quarks, etc. be satisfied;an empirical possibility are composed of smaller particles,which in turn could be composite, and so on. (I present this argument in detail elsewhere on (1993), as an objection to van Inwagen's proposed restriction composition.) A second objection to P1 might be based on the fact that not everypair of cases can be connected by a continuous series. No manyobjects to a continuous series connects any case withfinitely case withinfinitely manyobjects, for example.31However,it would be implausible to claim that,for example, thejump fromfinitude between composition and its lack. to infinity makes the difference But rejecting P1 because of such jumps would require claiming
a mereological atom is the fusionof itsunit 290n the usual terminology, class; let us understand "continuous series connecting cases C1 and C2"

onlyone atom. "cases" involving as excluding 10,and Unger1979.Un30SeevanInwagen1990a,72-73,and chapter of all composite objects;he believesin ger does not denythe existence
structures-see 241-42. molecules and certain crystal 311 thankEarl Conee for thisobservation. 217

THEODORE SIDER

somethinglike this,for,nihilismaside, one would be sayingthat all cases of noncomposition are separated from all cases of composition by a barrierover which no continuous series can cross. Next let us consider the notion of an "abrupt cutoff' in a continuous series. By this,I mean a pair of adjacentcases in a continoccurs, but in uous series such that in one, composition definitely does not occur.The second premthe other,compositiondefinitely ise of myargumentcan then be stated as follows: in whether P2: In no continuous series is there an abrupt cutoff composition occurs. This seems intuitively compelling. Recall thatadjacent membersin a continuous series were said to be extremelysimilar in certain respects.By including more and more members in each continuclose to being ous series,adjacent memberscan be made arbitrarily exactlysimilar in those respects. Given this,it would be hard to nearly as hard as it would accept the existence of a sharp cutoff, be to reject the supervenience of composition on the relevantfacforexample, thatalthoughcertain tors.It would involveour saying, particlescompose a larger object, if one of the particleshad been displaced 0.0000001 nanometers,those particleswould have failed in the apto compose any object at all. Of course, sharp cutoffs plication of a predicate are not alwaysimplausible-consider the predicate 'are separated by exactlythree nanometers'. What I object to is a sharp cutoffin a continuous series of cases of composition.32

A possible objection to P2 would be based on preciselystateable on the regions of space that can possibly topological restrictions be occupied by a composite object. For example, one mightallow fusions only when the occupied region of space would be connected (that is, when any two points of the region are connectable by some continuous path withinthe region). But thiswould seem etc. More to rule out too many objects: galaxies, solar systems, importantly, under the classical physicsconception of matter,all macroscopic objects are discontinuous.While this is less clear on
32Those happy with the "epistemic" view of vagueness, according to of vague predicatesis simplydue to our which the apparent indeterminacy here, since theyalreadyaccept ignorance,maybe happywitha sharp cutoff similarcutoffs for predicates such as 'heap', 'bald', etc. See for example Williamson 1994.
218

FOUR-DIMENSIONALISM

a quantum-mechanicalpicture,we stillwould not want to say that there would be no macroscopic objects in a classical world.33 The final premise of the argumentis, I think,the most controversial: P3: In any putativecase of composition,eithercompositiondefdoes not occur. initelyoccurs, or composition definitely P1, P2, and P3 togetherimplythe desired conclusion. P1 requires we have a case of compothat if composition is not unrestricted, sition connected by a continuous series to a case of noncomposiin thisseries where comtion. By P3, there must be a sharp cutoff position abruptlyceases to occur; but this contradictsP2. It must be emphasized that this is not "'ust another Sorites." The correct solution to traditionalSoritesparadoxes willsurelyinvolvein some waythe claim thatthere is a region in which the relevantpredicate applies nor definitely ('is a heap', 'is bald', etc.) neitherdefinitely But thisis failsto apply.There will be a region of indeterminacy.34 just what P3 prohibits. like to clarify its I turn now to the defense of P3. I would first intended content. Recall that a "case" was defined as involvinga conceived, as class of objects. I here mean classes as traditionally be distinguished opposed to "fuzzyclasses." Classes musttherefore memfromtheirdescriptions, which mightnot sharplydistinguish not bers fromnonmembers.P3 pertains to the classes themselves, to their descriptions.Thus, indeterminacyof truthvalue in the of my sentence 'The class of molecules in the immediate vicinity withP3. In virtueof body has a fusion' would not be inconsistent itsvagueness,the subject termof thissentence failsto referuniquely to any one class. Also note that P3 isn't concerned with the nature of the resultingfusion,but only withits existence. Given a whetherit certain class of molecules, it maywell be indeterminate with has a fusionthatcounts as a person. But thisisn't inconsistent have a fusion that is a borderline P3, for the class may definitely case of a person.
33I thankJohn G. Bennett for helpful observationshere. Another sort of fusionswould be to classes thatare sets.This seems of precise restriction were produced I wouldn'treally but ifsome motivation to me unmotivated, mind the restriction. of the epistemicviewof vaguenesswould disagree here; see 34Defenders note 32. 219

THEODORE SIDER

Lewis's method for establishingP3 runs as follows.In virtueof the definitionof 'fusion' in termsof parthood, we can formulate the assertionthat a given class, C, has a fusion as follows: (F) There is some object, x, such that (i) everymember of C is part of x, and (ii) everypart of x shares a part in common withsome member of C. (F) can be indeterminate in truth value35(relativeto an assignment to 'C') only if it contains at least one term such that it is indeterminate in meaning among various precise alternatives, or "precisifications."For example, a precisification of a two-placepredicate would be a two-placerelation such that it is alwaysdeterminate, given a pair of objects,whetherit holds betweenthem.It is difficult to see what the precisifications of logical terms,or the predicates 'is a member of and 'part of, mightbe. So (F) cannot be indeterminatein truthvalue. Lewis's justificationof P3 is weakest,I think,in its assumption that 'is part of cannot be a source of vagueness in truthvalue.36 His reason is that it is difficult of to see what the precisifications 'part of mightbe. But perhaps thisis due to these precisifications not being easily stateable in natural language. Some terms,such as the term 'is bald', seem to have easily stateable precisifications, namely,propertiesexpressed bypredicatesof the form'has a head withless than n hairs'.37 But other predicates are different. Surely there are or could be sentences of the form "ax is a person," or is a table," withpreciselyreferring that "13 singulartermsoaand I3, are indeterminatein truthvalue. But neither 'person' nor 'table' So we should be wary seems to have easilystateableprecisifications. of concluding that a predicate cannot be a source of vagueness fromthe fact thatwe can't thinkof what its precisifications might be. 35Notice thatthereare possible sourcesof truth valuegapsotherthan or failedpresupposition; I'll ignorethesein suchas ambiguity vagueness, the present discussion. Noticethatin ruling out 'partof as a sourceof vagueness, Lewisis of in ascriptionsof parthood, forascriptions not rulingout all vagueness 'theoutback')that are contain terms parthood may singular (for example, in which refer to. (F), however, conindeterminate objectthey apparently tainsno vaguesingular terms. is in doubt, of'head' in this thevagueness 37Even case,stateability given and 'hair'.
220

FOUR-DIMENSIONALISM

Fortunately, P3 may be supported withoutmaking any assumptions about parthood, for if there were vagueness in whether a certain class had a fusion,then there would be vagueness in how many concrete objects would exist,which is impossible. Let us stipulatively define concrete objects as those which do notfitinto any of the kinds on the followinglist: * sets and classes * numbers * propertiesand relations v universalsand tropes * possible worldsand situations If I've missed any "abstract" entitiesthat you believe in, feel free to update the list. Suppose now for reductio that P3 is false-that is, that there can sometimesbe vagueness in whethera given class has a fusion. In such a case, imagine counting all the concrete objects in theworld. One would need to include all the objects in the class in question, but it would be indeterminatewhether to include another entity: the fusionof the class. Now surelyifP3 can be violated, then it could be violated in a world withonly finitely many concrete objects. But consider what we may call "numerical sentences"-sentences assertingthe existence of particularfinite numbers of concrete objects. A numerical sentence assertingthat there are exactlytwo concrete objects,for example, looks like this (where the predicate 'Cx' means 'x is concrete'): 3x3y[Cx & Cy & x7y & Vz(Cz
-

[x=z V y=Z])]]

If it is indeterminatehow many concrete objects there are in a world withonly finitely many concrete objects, then some numerfalse. true nor definitely ical sentence must be neither definitely But numerical sentences contain only logical termsand the predicate 'is concrete'. The latterpredicate presumablyhas precise application conditions since it was simplydefined by the list given above, which consistsof predicates expressingfundamentalontological kinds that do not admit of borderline cases. And even if one of the members of the listis ill defined or vague in some way, then the vagueness is presumablyof a kind not relevant to my argument:any way of eliminatingthe vagueness would sufficefor present purposes. As for the logical terms,I accept Lewis's view that theycannot be a source of vagueness either.More carefully, 221

THEODORE SIDER

by sayingthat logic cannot be a source of vagueness, I mean that any sentence containing only logical expressions and predicates with no borderline cases (such as 'is concrete') must be either false. definitely true or definitely compelling;indeed, The viewthatlogic is non-vagueis extremely of preone would be inclined to cite logical concepts as examples cision. Consider the logical concepts case by case: boolean conAt the veryleast, in no case is and identity. nectives,quantifiers, in the way that there is for termslike there evidentindeterminacy sometimes is admittedly quantification 'bald', 'heap', etc. Restricted be quantifying can be vague: one could vague, for the restriction for example. But only over objects in one's "immediate vicinity," are intended to be entirely in the present context,the quantifiers unrestricted, since the issue in thissection is whetherforany class, sense a fusion. There are those there is in the most unrestricted who say that objecthood itselfis vague, and thus that even an unrestricted quantifiercan be a source of vagueness. I find this doctrine obscure, though I admit that I have no argumentagainst it. sentences can clearlyhave vague truthconidentity As for identity, ditions when they have singular terms that are indeterminatein reference-"Michael Jordan is identical to the most popular human being," forexample. But the onlysingulartermsat issue here which are not indeterminate are variables relativeto assignments, in reference.There are those who say that even withoutindeterminate singular terms,identity ascriptionscan be vague in truth value, despite Gareth Evans's (1978) argument to the contrary. Again, I find this doctrine obscure, but I have nothing to add to the literatureon Evans's argument. The argument for restrictedcomposition, we have seen, leans most heavilyon P3, which in turn restson the view thatlogic, and are nonin particular unrestrictedquantification and identity, vague. While this view is an attractiveone, I have had nothing to say in its defense. The presentargument,therefore, substantive should be taken as showing that anyone who accepts that logic is componon-vague must also accept the principle of unrestricted sition. In virtue of the parallel argument I will constructin the next two sections, everyone who shares this assumption about vagueness must also accept four-dimensionalism.

222

FOUR-DIMENSIONALISM

and tensed parthood questions 3.2 Composition The argumentof the previous section concerned the question of when a given class has a fusion, where 'fusion' was understood is in quesBut when the truthof four-dimensionalism atemporally. relativetalk sake I have advocated temporally tion, for neutrality's of parthood. If parthood is temporallyrelative,then so is the revariouscoma fusionof.This requires us to distinguish lation being position questions.38 The simplestquestion is that of when a given class has a fusion at a given time. But we are also interestedin what we mightcall classes at "cross-time"fusions-things thatare fusionsof different times. These are objects that gain and lose parts. One different concept of cross-timesummation may be introduced as follows. Let's use the term "assignment"forany (possiblypartial) function that takes one or more timesas argumentsand assignsnon-empty classes of objects that exist at those times as values; and let's say fusion("D-fusion," for short) of an that an object x is a diachronic of J(t). f ifffor everyt in f's domain, x is a fusion-at-t assignment For example, consider any two times at which I exist,and let f be a functionwith just those two timesin its domain, which assignsto each of those times the class of subatomic particles that are part of me then. I am a D-fusionof f since at each of the two times,I am a fusion of the correspondingclass of subatomic particles. A second question of composition,then,is the question of when a given assignmenthas a D-fusion:given various timesand various objects corresponding to each, under what conditions will there be some object that at the various times is composed of the corA third question would be that of the conresponding objects?39 ditions under which there would be such an object that existed onlyat the specified times. This is the question of when a given D-fusionof an where a minimal D-fusion, assignmenthas a minimal assignmentis a D-fusionof thatassignmentthatexistsonlyat times D-fusion of the in the assignment'sdomain. I am not a minimal f mentioned above, because I exist at timesother than assignment the two times in its domain. To get an assignmentof which I am a minimal D-fusion,simplyextend f into a functionthatassignsto
38See Simons 1987, 183ff.and Thomson 1983, 216-17. 39Thomsondiscusses this question (1983, 217). 223

THEODORE SIDER

any other time at which I exist the class of subatomicparticlesthat are part of me then. sense, a minimal D-fusionof some objects at varIn an intuitive ious times consists of those objects at those times, and nothing more. Though it has required some machineryto state,the queshave minimalD-fusionsis farfrombeing tion of which assignments remote and technical. Indeed, we can restatethis question in the do obfashion: Underwhatconditions followingwoollyyet satisfying Hof the a model we make and cease to exist?Suppose jects begin shaped part of Stonehenge out of three toyblocks, bl, b2, and b3, by placing one on top of two of the others at time t1; suppose we separate the blocks a few minutes later at t2. Is there something that we brought into existence at the firsttime and destroyedat the second? This is the question of whethera certain assignment has a minimal D-fusion-namely, the assignmentthat assigns the to everytime between t1 and t2. class {b1,b2,b3} forfour-dimensionalism vagueness from 3.3 The argument Under what conditionsdoes a given assignmenthave a minimalDfusion? I say that all assignmentshave minimal D-fusions,relying on an argument parallel to the argument for unrestrictedcomin minimal D-fusionswould require a cutoff Restricting position.40 as composicases. Just of similar some continuous series pairwise tion can never be vague, neither can minimal D-fusion. So the cutoffwould need to be abrupt, which is implausible. The argument may be preciselyformulatedas follows: P1': If not everyassignmenthas a minimal D-fusion,then there must be a pair of cases connected by a "continuous series" such thatin one, minimalD-fusionoccurs,but in the other, minimal D-fusiondoes not occur in whether P2': In no continuous seriesis therean abrupt cutoff minimal D-fusionoccurs P3': In any putativecase of minimalD-fusion,eitherminimalDdoes occurs,or minimalD-fusiondefinitely fusiondefinitely not occur
conbetweenmyargumentand arguments 40Thereare some similarities tained in Quine 1981 (10) and Heller 1990 (chap. 2, sec. 9). 224

FOUR-DIMENSIONALISM

The notion of a "case" must be adjusted in the obvious way.A "continuous series of cases" will now varyin all respects thought to be relevant to whether a given assignment has a minimal Dfusion. These respects mightinclude spatial adjacency,qualitative and causal relations at the various times in the assignsimilarity, ment,and also the beginning and cessation of these factorsat various timesof the assignment. The justificationof premise P1' is just like that for P1. As for in a continuous series of cases of minimalDP2', an abrupt cutoff similarin termsof spatial fusion-a pair of cases that are extremely adjacency, causal relations, etc., but which definitelydifferin whetherminimal D-fusionoccurs-seems as implausible as such a cutoffin a series of cases of composition. An objector might atby accepting a sort of mereologtemptto secure an abrupt cutoff nothing existsbut ical essentialismaccording to which,intuitively, mereological sums, which have their parts essentiallyand exist as long as those parts exist. If you accept this view then you can refor you can say that an asstrictminimal D-fusionsnon-vaguely, when and onlywhen it signmenthas a minimal D-fusion,roughly, longest assignmentfor a given fixed class of oW is the temporally jects.41 'the idea is that mereological fusionsof objects "automatiretain cally" come into existencewhen theirpartsdo, automatically go out of existence when any those same parts,and automatically of those parts go out of existence. So if you don't mind this sort you needn't fear my argument.I do of mereological essentialism, mind it, for it entails that nothing ever survivesthe loss of a part. on regions of space can provide Justas topological restrictions on composition (although I find them unmoprecise restrictions on regions of time can provide tivated), topological restrictions on minimal D-composition.I would reject the precise restrictions to continuous segmentsof time on the grounds thatwe restriction whereS1 and S2 are setsof objectsthatexistat times 41Less roughly: t, iffevery and t2,respectively, say thatpairs (tl, S1) and (t2, S2) are equivalent some member of S2, and every of any member of SI overlaps-at-t2 part-at-tl some member of S1. The idea of any member of S2 overlaps-at-tl part-at-t2 is thatS1 and S2 contain,ifnot exactlythe same members,at least the same is that an asThe non-vague restriction stuff, just divided up differently. signmentf has a minimal D-fusioniff f is a maximal equivalence-interrelated assignment;thatis (construing f as a class of pairs), iff(i) everytwo pairs in f are equivalent,and (ii) if (t, S) is equivalent to some member of member is t. f then it is equivalent to some member offwhose first
225

THEODORE SIDER

want to accept some objects withdiscontinuouslives (forexample, Hirsch's example of a watch that is taken apart for repairs and then reassembled), but othersmay disagree. One could precisely restrictminimal D-fusions by disallowing instantaneous objects; only objects that occupy an extended intervalwould be allowed. I would regard this restriction as unmotivated,but anyone who accepts either of these restrictions on minimal D-fusionsmay revise accordingly, formyargument mystatement of four-dimensionalism may easilybe adapted to supunrestricted four-dimensionalism for: four-dimensionalism while allowing for the restricport restricted tions on minimal D-fusions.On the restricted version of four-dimensionalismthere will be neither temporally discontinuoustemporal segmentsnor instantaneoustemporal parts,although there will be temporal segments of arbitrarily small duration. There is four-dimensionalism is fourlittleneed for a fighthere: restricted dimensionalismenough. whethercomMy argumentforP3 was thatifit is indeterminate position occurs then it will be indeterminatehow many objects thereare, which is impossible.I use a similarargumentto establish P3'. Indeterminacy in minimalD-compositionmightbe claimed in several situations.But in each case, I will argue, at some possible world there would result "count indeterminacy"-an indeterminacy in the finitenumber of concrete objects-which, as I argued above, is impossible, assuming that logic is not a possible source of vagueness. (Recall the distinction between existence-at and is indeterminacy in how many Count indeterminacy quantification. objects there are,not merelyin how many of the objects there are that existat some specified time. It is the formerthat I need to argue would result from indeterminacyin minimal D-fusion,berules out is the cause all that my assumption about logic directly former.) I distinguishfour situationsin which someone might claim inin whetherminimal D-fusionoccurs: determinacy someobjects have a fusionat a given (i) Indeterminacy as to whether This would are moderately scattered at thattime. time, say,becausethey For consider a possible world conresult in count indeterminacy. taining some finitenumber of quarks that at all times are greatly
42See Hirsch 1982, 22ff., as well as the whole of chapter 1, on different senses of "continuity." 226

FOUR-DIMENSIONALISM

scattered,except for a single time at which they are moderately scattered.Moderate scatteringwas alleged to result in indeterminacy, so it should have that result in the present scenario at the one time when the objects are moderatelyscattered.But the objects are much more scatteredat other times,so the resultwould in how many objects exist at the world in quesbe indeterminacy tion: there is one more object in the world depending on whether would apply the quarks have a fusionat thattime. (Similar remarks if 'scattered'in thisparagraphwere replaced byvariousotherpredicates deemed relevant to the question of whether a class has a fusion at a given time.) an object thatis a fusionat t ofcertain in whether (ii) Indeterminacy time, t', of is identical to an object thatis a fusionat someother particles someother This, too, would resultin count indeterminacy. particles. Suppose I undergo amnesia in such a way thatwe feel indeterminacy in whether "Young Man Ted is identical to Old Man Sider" is true. Presumablywe will want to say the same thing about this manyconcrete things. case if it occurs in a world withonly finitely in whethera if indeterminacy do have in this we really world, But certainassignmenthas a minimalD-fusion(say,one thatassignsto times before and afteramnesia all my parts at those times), then in the count of the concrete objects therewillresultindeterminacy there, for if the identityholds then there will be one less object than if the identity does not hold. in when an objectbeginsto exist.Again, this (iii) Indeterminacy Suppose, for example, that would result in count indeterminacy. in some case, C, it is indeterminatewhen a certain statue comes into existence. Consider next a case much like C, but in which (a) manyconcrete thingsexist,and (b) the molecules that onlyfinitely the timeat which would make up the statueare all annihilated after exist. Then it will be inthe statue is alleged to indeterminately determinatewhether the statue exists at all, and hence indeterminate how many thingsthere are at the world in question. This case is simceases existing. in whenan object (iv) Indeterminacy ilar to the previous case. We have seen, then, that if any of the offeredreasons for there being indeterminacyin whether a given class has a minimal Dfusion are genuine, then we would have to accept that at some many concrete things,it is indetermipossible world with finitely nate how many concrete objects exist there. But then therewould
227

THEODORE SIDER

true nor defibe a numerical sentence that is neither definitely can issue from nitely false.Assumingas I am thatno indeterminacy logic, this is impossible. So P3' is true: a given assignmentmust lack a minimal D-fusion.This is either definitely have or definitely not to say that the phenomena adduced in (i)-(iv) are not genuine; they simplymust be understood in some way not implying in minimal D-fusion.One waywould be as follows: indeterminacy on evis due to indeterminaterestrictions (i) The indeterminacy we do not quantifyover all the erydayquantification.Typically, objects that there are, but only over fusionsof objects that aren't too scattered;ifobjects are borderlinescatteredat some time,they stilldefinitely have fusionsat those times,but we have a borderline resistance to admittingthose fusions into an everydaydomain of quantification.(ii) This is a case involvingthree objects. Object 1 begins around the time of mybirthand ends at the amnesia, Object 2 begins at amnesia and lasts until my death, and Object 3 lasts throughoutthis time interval.The name 'Young Man Ted' is indeterminatein reference between Objects 1 and 3; the name 'Old Mati Sider' is indeterminatebetween Objects 2 and 3; hence the identitysentence is indeterminatein truthvalue. (iii) There in when theybegin to exist; are manyobjects involved,which differ the term 'the statue' is indeterminatein referenceamong them; hence the sentence 'The statue begins to exist at t' will be indeterminatein truthvalue for certain values of 't'. (iv) is similarto (iii). Pi', P2', and P3' jointlyimply: (U) Everyassignmenthas a minimal D-fusion. The But (U) is a powerfulclaim,forit entailsfour-dimensionalism! centralfour-dimensionalist claim, recall, is the Thesis of Temporal Locality: non-overforany object x, and forany non-empty, Necessarily, lapping sets of times T1 and T2 whose union is the time span of x, there are two objects xl and x2, such that (i) xl and x have the same parts at everytime in T1, (ii) x2 and x have the same parts at everytime in T2, and (iii) the time span of xl = T1, while the time span of x2 = T2 Let x, T1, and T2 be as described; xl is obtained by applying (U) to the assignmentthat assigns x's unit set to all and only t in T,; 228

FOUR-DIMENSIONALISM

The Thesis of Temporal Locality,then,is true. So similarly forx2.43 including the claim thatan object must too are all its entailments, have a temporal part at everymoment at which it exists.We have is true. resolved our dispute: four-dimensionalism University ofRochester
References in PhilosStudies Adams, Robert M. 1986. "Time and Thisness." In Midwest 315-29. Minneapolis: UniophyX1,ed. French, Uehling, and Wettstein, of Minnesota. versity and Cause:Essays Through Time." In Time D. M. 1980. "Identity Armstrong, ed. P. van Inwagen, 67-78. Boston: Reidel. toRichardTaylor, Presented Quarterly J. 1984. "Spatial and Temporal Parts." Philosophical Butterfield, 35:32-44. La Salle: Open Court. and Object. Chisholm,Roderick. 1976. Person XI: Studies in Philosophy Dau, Paolo. 1986. "Part-TimeObjects." In Midwest Studiesin Essentialism, ed. P. A. French, T. E. Uehling, and H. K. Wettof Minnesota. stein,459-74. Minneapolis: University 38:208. Evans, Gareth. 1978. "Can There Be Vague Objects?" Analysis Fine, Kit, 1994. "Compounds and Aggregates."Nofis28:137-58. Geach, Peter. 1972. "Some Problems about Time." In his Logic Matters, of CaliforniaPress. 302-18. Berkeley:University ofPhilosophy Journal Graham, George. 1977. "Persons and Time." Southern 15:308-15. Haslanger, Sally. 1994. "Humean Supervenience and Enduring Things." Journal 72:339-59. Australasian ofPhilosophy JourHeller, Mark. 1993. "Varietiesof Four Dimensionalism." Australasian 71:47-59. nal ofPhilosophy Research 52: and Phenomenological 1992. "Things Change." Philosophy 695-704. 43Proof: (U) tells us that some object xi is a minimal D-fusionof the assignment, f withdomain T., which assigns {x} to everymember of TI. The time span of xl = TI, since (a) xl existsonly at timesin f 's domain, and no time outside of T. is in f 's domain, and (b) x, containsx as a part and hence existsat everyteTI. Moreover,where t is any time in T1, xl is a fusion of {x} at t. We now show that xl and x have the same parts at t: (i) Let y be part of x at t; x is part of xl at t since xl is a fusionof {x} at t; but then y is part of xl at t. (ii) Let y be part of xl at t, and suppose for reductio thaty isn't part of x at t. By (PO) (see note 9), y has a part,z, at t thatx doesn't overlap at t. Since x, is a fusionof {x} at t, x overlapsevery part of xl at t. But z is part of xl at t. Contradiction. the resultfollows. Withoutloss of generality, 229

THEODORE SIDER HunksofMatObjects: FourDimensional ofPhysical . 1990. TheOntology Press. Cambridge: Cambridge University ter. Objects." Philosophical Dimensional of Four Parts . 1984. "Temporal Studies 46:323-34. Press. Oxford: Oxford University ofIdentity. Hirsh,Eli. 1982. The Concept ofthe Hughes, C. 1985. "Is a Thing Justthe Sum of Its Parts?" Proceedings Aristotelian Society: 213-33. Kazmi, Ali Akhtar.1990. "Parthood and Persistence." CanadianJournalof supp. vol. 16:227-50. Philosophy, Leonard, Henry,and Nelson Goodman. 1940. "The Calculus of Individuals Logic5:45-55. and Its Uses." JournalofSymbolic Oxford: Basil Blackwell. ofWorlds. Lewis, David. 1986a. On thePlurality Press. vol. 2. Oxford: Oxford University Papers, 1986b. Philosophical 62: Journalof Philosophy 1984. "Putnam's Paradox." Australasian 221-36. Papers, vol. 1, 55In his Philosophical 1983. "Survivaland Identity." Press. 77. Oxford: Oxford University Studies. Markosian,Ned. Forthcoming."Brutal Composition." Philosophical Press. Mellor,D. H. 1981. Real Time.Cambridge: Cambridge University of Enduring and PerMerricks,Trenton. 1995. "On the Incompatibility during Entities." Mind 104:523-31. 91: JournalofPhilosophy 1994. "Endurance and Indiscernibility." 165-84. of California Berkeley:University John, ed. 1975. PersonalIdentity. Perry, Press. Prior,A. N. 1968. "Changes in Events and Changes in Things." In his Press,Clarchap. 1. Oxford: Oxford University Paperson Timeand Tense, endon Press. and Things. Cambridge: Harvard University Quine, W. V. 0. 1981. Theories Press. Studies in the . 1976. "WhitherPhysicalObjects." In Boston Philosophy Lakatos,ed. R. S. Cohen, P. K. ofImre vol. 39: Essaysin Memory ofScience, 497-504. Boston: Reidel. Feyerabend,and M. W. Wartofsky, Journalof Sider, Theodore. 1996a. "All the World's a Stage." Australasian 74:433-53. Philosophy Un. 1996b. "Merrickson Perdurance/EnduranceIncompatibility." published. of Gunk." Analysis 53:2851993. "Van Inwagen and the Possibility 89. Oxford: Oxford University in Ontology. Simons, Peter. 1987. Parts:A Study Press. AcrossTime." JourJarvis.1983. "Parthood and Identity Thomson, Judith nal ofPhilosophy 80:201-20. 230

FOUR-DIMENSIONALISM

Unger, Peter. 1979. "I Do Not Exist." In Perception and Identity: EssaysPresented to AJ AyerwithHis Repliesto Them, ed. G.F. Macdonald, 235-51. New York:Macmillan. van Inwagen,Peter. 1990a. Material Beings. Ithaca: Cornell University Press. 1990b. "Four-DimensionalObjects." Nofis24:245-55. . 1981. "The Doctrine of Arbitrary Undetached Parts." Pacific PhilosophicalQuarterly 62:123-37. Wiggins,David. 1980. Sameness and Substance. Oxford: Oxford University Press. Essential Depen. 1979. "Mereological Essentialism:Asymmetrical dence and the Nature of Continuants." In Essayson the Philosophy ofRoderick Chisholm, ed. E. Sosa, 297-315. Amsterdam:Rodopi. London: Routledge. Williamson,Timothy.1994. Vagueness.

231