Constitution Is Not Identity

Mind Association
Constitution Is Not Identity Author(s): Mark Johnston Source: Mind, New Series, Vol. 101, No. 401 (Jan., 1992), pp. 89-105 Published by: Oxford University Press on behalf of the Mind Association Stable URL: http://www.jstor.org/stable/2254121 . Accessed: 17/04/2013 17:03
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.
Oxford University Press and Mind Association are collaborating with JSTOR to digitize, preserve and extend access to Mind.
http://www.jstor.org
This content downloaded from 159.149.103.9 on Wed, 17 Apr 2013 17:03:09 PM All use subject to JSTOR Terms and Conditions
ConstitutionIs Not Identity

MARK JOHNSTON
I
Suppose that a statue of Goliath is made by fusing together two appropriately shapedpieces of clay and thataftera few minutes,the artisan,frustrated with his work, dissolves the statuein a solvent which destroysclay and statuealike. Then a naturalthing to say is thatthe careersof the statueandthe lump or piece of clay which made it up are entirely coincident. The statue and the piece of clay came into being at the same time and ceased to be at the same time. Throughouttheir respective careers,the piece of clay constitutedthe statue. Had the artisandespairedonly of the armsand calves of Goliathand dissolved only them,replacingthem with new pieces of appropriately molded clay, then we should say that distinct but not wholly distinct pieces of clay constituted the statueof Goliathover its lifetime. In this second case we naturallyconclude thatthe statueis not absolutelyidentical with the whole piece of clay which originally constitutedit, since the piece arguablydid not survive the dissolving of significantpartsof it, while the statue clearly did survive the dissolving; as is evidenced by the fact that the statuehad new arms and calves attachedto it. So also, it seems naturalto conclude that even in the first case in which the originalpiece of clay constitutedthe statuethroughout its entirecareer,the statue is not absolutelyidenticalwith the clay, since the statuecould have survivedcertain changes which the piece of clay would not have survived, e.g. the changes describedin the second case. Philosophers have gone to some lengths to resist this last conclusion. Thus David Lewis, Alan Gibbard, Anil GuptaandDenis Robinsonall allege thatsomething special aboutmodal predicationinvalidatesthe argument to non-identityin the case of complete coincidence.' Concentrating on Lewis's way of puttingthe point, since it fits neatly into a familiarsystematic way of thinkingof modality, the situationis supposed&to be as follows.2 If the term "Lumpl"names the piece of clay which makes up the statueGoliath, the aim is to defend (1) In the case of complete coincidence, Lumpl = Goliathwhile allowing (2) Lumplcould not have survivedthe loss of those of its partswhich made up the arms and calves of Goliath. Goliathcould have.
2
' See Lewis (1971), Gibbard(1975), Gupta(1980) and Robinson (1982). The remarksthatfollow are adaptedfrom Lewis (1971). ? Oxford University Press 1992
Mind, Vol. 101 . 401. January 1992
90
MarkJohnston
How then are we to avoid the following contradiction,got by substitutingin (2) on the basis of (1)? (3) Lumplcould not have survivedthe loss of those of its partswhich made up the arms and calves of Goliath.Lumpl could have. Lewis's answeris essentially this: The position held by "Goliath"in (2), even if de re, i.e. outside of the scope of the modal "could have", is not referentially transparent, i.e. does not supportthe substitutionof codesignative names salva veritate.This is because "Goliath" as a name of a statueis performingmore than a referentialfunction in (2). "Goliath"not only picks out the statue but it also, perhapswith the cooperationof context, indicates a particularcounterpart relation or set of necessary and sufficient qualitativeconditions for tracingGoliath across the worlds-if you like, the "statue"counterpart relation.3This relation determinesfor Goliath a set of otherworldly,relevantly similar individuals, or counterparts,which representthe possibilities for Goliath. We have a different set of counterpartsdeterminedif we employ the "lump of matter"counterpart relation indicated by the name "Lumpl".Since, on the view in question, what could have happenedto a thing is what happensto at least one of its counterparts, the truthvalues of the modal predicationin (2) are sensitive to which counterpart relationor principleof cross-worldtracingis invokedby the namesused (perhaps in conjunctionwith context).Thatis why we have (2) trueand (3) false. The substitutionof codesignativenames altersthe tracingprinciplesin termsof which the relevantmodal predicationsare to be evaluated. It is instructiveto investigatethose analogousmoves which in the case of temporal predicationcould be employed to defend (1') In the case of only partialcoincidence, Lumpl=Goliath. The obvious problemwith (1') is (2') Lumpl will not survive the loss of those of its partswhich make up the arms and calves of Goliath.Goliath will. In conjunctionwith (1'), this seems to imply a contradiction,namely (3') Lumpl will not survive the loss of those of its parts which make up the armsand calves of Goliath. Lumpl will. There are three slightly differentways of understanding temporalpredicationso as to make (1') and (2') truebut avoid the inferenceto the contradiction apparent in (3'). The first way takes "Goliath"and "Lumpl"to be names of continuants, and claims that (2') is truebecause it is properlyread as the loss of those of its parts which (2*) Lumpl will not survive-qua-Lumpl make up the arms and calves of Goliath. Goliath will survive-quaGoliath.
3 For counterpart relatheory see Lewis (1968). Lewis discusses multiple counterpart tions in Lewis (1986), ?4.5.
Constitutionis Not Identity
91
where something survives-qua-Lumpl some event just in case Lumpl is around before and after the event, and something survives-qua-Goliathsome event if Goliath is aroundbefore and afterthe event. From (1') and (2*) we derive (3*) Lumpl will not survive-qua-Lumpl the loss of those of its parts which make up the arms and calves of Goliath. Lumpl will survivequa-Goliath. (3*)is true on the intended construal of "x survives-qua-y".However, the unattractiveartificialityof this suggestion is immediatelyobvious. Survivingsome event is plausiblytakento consist in being aroundbefore and afterthe event, but "Goliathsurvives-qua-Lumpl" (or "Goliath-qua-lump survives")just means that Lumpl survives, and "Lumplsurvives-qua-Goliath" (or "Lumpl-qua-statue survives") just means that Goliath survives. Better then to avoid this kind of "qua" talk, since it only serves to mask paradox if a's surviving-qua-b has nothing essentially to do with a's continuingto be around.The real force of (2') is not capturedby this first proposal. Mattersare not much differenton the second proposal,which takes "Goliath" and "Lumpl"to be names of continuants now explicitly understood as fourdimensionalsums of temporalstages unitedby appropriate gen-identitycriteria, i.e., criteriafor aggregatingthe short-livedstages into longer-lived continuants. Then it is said that for the continuantx to survive-qua-Lumpl or to survive-qualump some event e is for there to exist, after e, lump of clay stages which stand in the relation of being stages of the same lump of clay to stages of x existing before e. Mutatismutandisfor the continuant x to survive-qua-Goliath (or to survive-qua-statue or qua-statue to survive).Now we construe(2') as sayingthat(2*) andreplace(3') by (3*). The inferencefrom(1') and(2') to (3*) is unproblematic. But once again we have a quite artificialsense of "survival"at work:x's surviving-qua-ysome event e has nothingessentially to do with x being aroundafter the event e. To say thatx exists in world w is to say no more thanthatthe abstract of the way the world might have been thatis w representsa possirepresentation bility for x. This leaves it open just how this possibility is representedby w, e.g., or by havingx's counterparts by havingx do the representing do it. But to say that x actually survives e is to say something that requiresx to exist after e. Mere descendantsof x won't do.4 The third theory of temporalpredicationis the closest analogue of Lewis's multiplecounterpart and "Goliath" each denote a theory and has it that "Lumpl" stage, and in fact the sanie stage, a stage existing at the periodduringwhich (1')4 Some will say thatI have illegitimatelysupposedthatwe can always pick out a continuantx, whereasin this context one can only pick out x-qua-statueandx-qua-clay.I find this very implausible.To see its implausibility,imagine that we view the history of the statue and the clay from a laterdate. Why can't we then introducethe name "Goliath"to denotethe continuantwith the more inclusive-history-i.e., the historyof the statue,which continued on after the change of parts-and introducethe name "Lumpl"to denote the continuant with the less inclusive history-the history which came to an end with the is irreducible change of parts.The claim thatqua-prediction may look plausiblewhen contemplatingfutureoccurrencesbut its plausibilityvanishes when we contemplatehistories from the perspectiveof the remembrance of things past.
92
Mark Johnston
(3') are uttered.But the theoryhas it thatthese names do more semanticallythan merely denote. They carrywith them, perhapsvia the associatednominal"piece of clay" and count noun "statue",distinct gen-identity criteria, i.e. criteriafor aggregatingmomentaryor short-livedstages into longer-livedcontinuants.Necessary and sufficient conditions for being a stage of the same persistingpiece of and clay as Lumplis a stage of are somehow associatedwith the name "Lumpl", necessaryand sufficientconditionsfor being a stage of the same persistingstatue Since thatGoliathis a stage of are somehow associatedwith the name "Goliath". a stage will survive some change only if it is gen-identical with a stage which exists afterthe change, the questionas to whetherthe stage referredto twice over in (1') will survive some specific change is sensitive to which name is used in putting that question. Hence, although (1') and (2') are true while (3') is false, (3') does not follow from (1') and (2'). This theory of temporalpredicationalso breaksdown. For it to begin to work we must once again think of continuants as four-dimensional aggregates of stages. Then it is naturalto say that continuantsundergoor survivechanges like coming to be constituted by different matter, while their stages succeed each other through such changes. The change in question is the replacementof the armsandlegs of Goliathandhence of muchof the matterof Lumpl.On the theory in question "Lumpl"and "Goliath"are supposed to denote a pre-replacement stage sharedby the statue and the piece of clay. But then it follows that Goliath (the pre-replacement stage) does not survivethe changedescribedin (2') and (3') any more than Lumpl does. The continuantstatuethat includes Goliath literally survives but, on the thirdtheory, this continuantis not Goliath. On that theory, "Goliath"names a short-livedstage. So the thirdtheory of temporalpredication is just like the first two theories in failing to capturethe force of (2'). Despite the similarities between the cases of complete and partial coincide re temporal dence, thereis little to be said for defending(1') by understanding predicationas introducing referentialopacity.Yet in ordinarylanguagethe use of modal and temporalpredicationseems strikinglysimilar.(They shade into each other when the topic is which possibility is likely to occur next.) So we should inquireinto the claim which motivates the construalof de re modal predication as introducingreferentialopacity. This is the claim that (1) is true, more generally the claim that a materialobject is identicalwith the matterwhich constitutes it over its lifetime.
11
I know of only two argumentsfor (1) with any primafacie plausibility.The first is an argumentfrom mereology, allegedly the logic of the part/wholerelation.
is NotIdentity 93 Constitution Mereology employs the primitive predicate "O(x,y)"where the intended construalof this predicateis thatx and y "overlapin content"in the broadest"logical" or topic neutral sense of these terms. (As we shall see, it is an important questionjust what this sense is.) Given this notion of overlap,the logical or topic neutralnotion of one thing being a partof anothermay be defined thus
(4) P(x,y)
*-*
Vz(O(z,x)
->
O(z,y))
Mereology is interestingfor ourpurposesbecause, dependingon the details of its formulation,mereology contains as an axiom or a theorem
(5) x=y
*-*
Vz(P(z,x)
*-*
P(z,y))
Now we seem to have a powerful tool for arguingto identities. Exhaustively enumeratethe "parts"(in the broadesttopic neutralsense) of x and of y. If they have just the same parts"they"are identical.5So the mereologist might arguein favor of (1) as follows: clearly in the case where Lumpl and Goliath are coincitheirentire careersthey have just the same parts-there is nothdent throughout ing that is a part of Lumpl that is not a partof Goliath nor vice versa. After all, isn't this what we are saying when we say that throughouttheir careers all of Lumpl constitutes Goliath and Goliath is constitutedonly by Lumpl. But then, using (5), we must conclude thatGoliath=Lumpl. So long as we think of continuantsas four-dimensionalsums of stages, the does not generalizeto encompass(1') understoodas a claim aboutconargument tinuants.In the case of partialcoincidence the four-dimensionalstatue Goliath piece of clay Lumpl. will have differentmaterialpartsfrom the four-dimensional After the change of head and calves there will be partsof Goliath which are not partsof Lumpl. for (1) standup to scrutiny?I thinknot. Indeed,I believe it Does the argument theirentirecareersGoliathis Throughout simply equivocateson the term"part". wholly constituted by Lumpl. One thing wholly constitutes another iff every materialpartof the first standsin the ancestralof being a materialpartof to some material part of the second.6 At most, then, constitution implies sameness of materialparts.But we cannotplausibly construe(5) as licensing claims of identity simply on the basis of samenessof materialparts.For supposewe have a plasticene pot. We squashthe plasticene, therebydestroyingthe pot. We then use the plasticene-all of it-to make a bust of Napoleon. Clearlythe pot is not identical with the bust, even thoughthey have the same materialparts.Someone might say thatthe troublehere is thatthe pot and the plasticene, althoughmade of the same matter,were made of it at differenttimes, so that we get identityjust when we have x andy made of the same matterover just the same rangeof times. But this conditionfor identityis too close a claim to the claim that(1), i.e., thatin the case of complete coincidence, Goliath=Lumpl,to provide an argumentfor (1). AnyFor examples of such uses of mereology see Thomson (1983). wording is to accommodatethose who might wish to worry The slightly roundabout whether for example the subatomic parts of the clay are really material parts of the statue-maybe the statues'smaterialpartsare only partslike the clay, the legs, the torso, the arms, and the head.
5
6
94
MarkJohnston
one who has doubtsabout(1) will doubtthe condition.Nor will it help to cast the condition in the language of materialtemporalparts,where a materialtemporal part is the content of a matter-occupied region of space-time, and materialcontinuantsare thoughtof as sums of such temporalparts.For then the argument for (1) which goes by way of
(5) x=y
*-*
Vz(P(z,x) e P(z,y))
interpreting "P(z,x)" as z is a temporalpartof x, and understanding Goliath and Lumpl as sums of temporalparts,will simply beg the question against someone who has doubtsabout (1). Such a person will rejectthe motivatingmetaphysical model of Goliathand Lumpl as four-dimensionalsums of temporalparts. If thereis to be a non-questionbegging argumentfor (1) from (5), the mereologist owes us an account of a topic neutralor logical notion of a partfor which (5) is true. Only then can he be said to be motivating(1) by means of (5), rather than appealingto a metaphysicalview of continuantswhich is plausible only if particular claims like (1) are true. But what is this logical or topic neutralnotion of a part, supposedly availableprior to conversion to metaphysicalviews about continuants? Unfortunately,mereologists are far from unanimouson this point. Thus, for example, in theiroriginalpaper"TheCalculusof Individualsand Its Uses" H.S. Leonardand Nelson Goodmanwrite in our interpretation [of mereology] parts and common parts need not necessarily be spatial parts.Thus in our applicationsof the calculus to philosophicalproblems,two concreteentities to be taken as [non-overlapping or] discretenot only have to be spatiallydiscrete,but also temporally discrete, discrete in color, etc. etc. (1940, pp. 46-7; my emphasis) The idea of objects being mereological sums of propertyparts with the explicit conditionthatpropertiesare in the first instanceparticular and not universalwas a leading motive in the eminently clear-headedmetaphysicsof D.C. Williams.7 David Lewis, also working within a mereological framework,opts for a more restrictivenotion of a propertypartwhen he writes It is quite easy to believe that a point particle divides into a few nonspatiotemporal partsin such a way thatone of them gives the particleits charge, anothergives it its mass and so on. But it is just absurdto think that a thing has (recurringor non-recurring)non-spatiotemporal parts for all its countless abundant properties.(1986, pp. 66-7) construalof the notionof partrelevantto mereology Elsewhere,an even narrower for the has set in. Thus JudithJarvisThomson,discussing a primafacie argument doctrineof temporalparts-an argumentwhich involves applyingmereology to a scatteredsum of parts which was temporarilyunscatteredand shaped in the form of a house-writes
7 Williams (1966) and (1986).
95
Even if thereis [suchan entity as house-shape]it certainlyis not literally partof any house. (1983, p. 201) Evidently there is much dispute about the metaphysicalparts of objects and of the crucial hence much dispute aboutthe detailed metaphysicalinterpretation mereologicalpredicate"O(x,y)". Adjudicatingthe dispute is not to the point here. For we are not interestedin implementinga metaphysicsbut only in evaluatinga mereological argumentfor the conclusion thatan object is identicalwith the matterwhich invariablyconstitutes it. In evaluating that argumentwe should not rely upon any substantive metaphysicalconception of the partsor ingredientsof entities. Ratherour question is whetherthereis any metaphysicallyneutralor logical conceptionof a part judgementsaboutconstitution which could be used in conjunctionwith ordinary and materialpartsto motivateclaims such as (1). For those who arecomfortablewith thinkingof second orderlogic as the logic of propertiesthe following may provide a way of thinking about the question. This principleof second orderlogic (6) x=y - VF(Fx e Fy) has a good claim to define the logical notion which is identity.From left to right we have Leibniz's Law, surely an a prioriprincipleif anythingis. From right to left we have a triviality secured by instantiation;among the propertieswhich x and y are assertedto have in common by the right hand side are identity with y and identity with x. Moreover, from (6) we can prove the familiar features of identity, viz., that it is reflexive, symmetric and transitive.So the second order definitionof identityhas a good claim to be the definition of identity. Now in standard mereology the following is the definition of identity:
(7) x=y
<-*
Vz(O(z,x)
<-*
O(z,y))
and So it seems thatif we want to give "O(x,y)"its properlogical interpretation, by means of

(4) P(x,y) <-- lz(O(z,x) --- O(z,y))
fix the logical interpretation of "P(x,y)",and therebyfind out what a logical part the mereois, we should consider this question:How then must we understand of identityis logical primitive"O(x,y)"so thatthe second ordercharacterization of identity? logically equivalentto the mereological characterization This quite determinatequestion is the question to face if we are to take mereof the logic of identity"i.e. a set of principles,includology as a "generalization of identity, which purportto have as much claim on us as ing a characterization (6) x=y -> F(Fx -> Fy) Our problem then is to find an appropriate pure logical construalof the mereological primitive"O(x,y)". the mereologicalprimitive"O(x,y)"as saying thatx is Clearly, if we interpret identicalwith y, then (7) and (6) will be logically equivalent.For then Vz(O(z,x)
*-*
O(z,y)) is logically equivalent to x=y.
96
Johnston Mark
If we interpret the mereologicalprimitive"O(x,y)"as saying thatx is not identical with y, then (7) and (6) will be logically equivalent.For then Vz(O(z,x) O(z,y)) is logically equivalentto x=y. not do to underIn orderto make (7) come otttlogically equivalentto (6) it "will stand"O(x,y)"as meaningthatx andy standin some relationR thatneitherguarantees identity nor guarantees distinctness. For then there will be no logical guaranteethatx=y just in case for every z, z standsin R to x if and only if z stands in R to y. observationis this: if we accept (6), the second order The simple but important the of identity,and we ask ourselves how we should understand characterization mereologicalprimitive"O(x,y)"so thatthe mereologicaldefinition
-
(7) x=y *-* Vz(O(z,x) *-* O(z,y)) has the same status as (6), the answer is that we should take "O(x,y)"either to guaranteethatx is identical with y or to guaranteethatx is not identical with y. Given this conclusion and (4) P(x,y) X- Vz(O(z,x) -4 O(z,y)) it follows that whicheverof these two ways we construe"O(x,y)","P(x,y)"will mean thatx is identicalwithy! The exercise has shown what most of us probably suspectedall along: there is availableno non-trivialnotion of a logical part.x is a logical part of y just in case x is y. Pure logic can offer no interestinginterpretation of the predicate"P(x,y)". Of course, the mereologist may accept this and yet still urge (7) on us as an extension of what we learnedin the logic class. (He may not be very interestedin securing the honorific "logic" for a principle in which he has complete confidence.) However, what should embarrassthe mereologist is that we do have a of identity, viz. the second ordercharacperfectly good logical characterization can either impeach or terization,whose status no mereological characterization approach. That status is importantlydifferentfrom the status of the metaphysicalviews about objects which would guaranteethe equivalence of (6) and (7). These two of identity would be equivalenton the assumptionthat objects characterizations Then (7) are sums of theirproperties,and hence have their propertiesas "parts". would have it thatx is identical with y just in case every sum of propertieswhich overlapswith x overlapswith y and vice versa. But the pictureof objects as sums of propertiesis neitherforced on us by logic nor capable of being the upshot of any empiricaldiscovery.In thatsense it is sheer metaphysics.Moreover,in order in a defence of the claim that to use (7) so interpreted (1) In the case of complete coincidence, Lumpl=Goliath we would have to accept the furthermetaphysical principle that sameness of materialor spatiotemporal partsis sufficient for sameness of propertyparts.
97
Obviously, all this is substantialmetaphysics, not formal logic. There is no metaphysicallyneutral,logical conception of a partwhich could be used in conjunction with ordinary judgements aboutmaterialpartsso as to motivate claims to merely such as (1). If that is so, then in motivatinga metaphysics,asaopposed implementingone, mereology is aboutas useful as Mariology. The metaphysicallyneutralguide to identityis just (6) (6) x=y -> F(Fx ->Fy) and in our original case of complete coincidence we have good reason to think that there is a propertythat holds of Goliathbut not of Lumpl:it is the property of being able to survive the loss of those of its materialpartswhich make up the armsand calves of Goliath.Alternatively,supposingLumplto be squashable,we have good reason to think that there is a propertythat holds of Lumplbut not of Goliath:it is the propertyof being able to survive being squashed.Appeals to (7) or (5) on behalf of (1) notwithstanding, there is good reason to think that (1) is a logical principle. false since (6), unlike (7) and (5), is uncontroversially Even so, some are unsatisfied by this use of (6) against (1) precisely because it depends upon exploiting modal or dispositional differences as between things with just the same categorical intrinsic properties.The second argumentfor (1) develops the worry that sameness of categorical intrinsics is sufficient for identity.
111
is This second argumentfor (1), thoughnot explicitly employed in the literature, one that seems to me especially worthyof respect. Sometimes in discussion it is said on behalf of (1) "How can you distinguishbetween Goliath and Lumpl on the basis of mere modal or counterfactual or dispositionaldifferences;after all, Goliathand Lumplare intrinsicallyexactly alike, how could they fail to be idenAs it stands,this goes too far, for thereis no reliableroadto identityfrom tical?"8 qualitativesimilarity.After all (6) x=y -> F(Fx ->Fy) is a theorem of second order logic only because the second order quantifier ranges over non-qualitative propertieslike identity with x. A fortiori,there is no utterlyreliableroadto identityfrom intrinsicqualitativesimilarity.But thereis a powerful point in favor of (1) to be made by exploiting the fact that Goliath and Lumpl seem to be intrinsicallyexactly alike, i.e. both are made up of exactly the same sort of clay, have exactly the same shape and texture, weigh exactly the same and so on. Takeany sort of thing F, then this principlewill be plausible:
8 Here, and in what follows, those who think of modal propertiesand dispositions as reto mean categoricaland intrinsic.I particularly intrinsicshould understand "intrinsic" just quoted. memberDenis Robinson urgingthe argument
98
MarkJohnston
(8) If y is a paradigmF andx is intrinsicallyexactly like y then x is an F.9 Now suppose for reductio ad absurdumthat Goliath and Lumpl are distinct. Goliath is a paradigmstatue of Goliath and Lumpl is intrinsicallyexactly like Goliath. By (8), Lumpl is a statueof Goliath.Two bizarreconsequences follow. First, an intolerabletype of co-occupancy:in the case in which Lumpl and Goliath are completely coincident,we have at least two statues of Goliath in the same theirentirecareers.At least two because we can invent a name place throughout for the piece of matterwhich constitutesLumpl, say "Lumpl*",and arguefirst thatLumpl*is distinctfrom the statuesLumpland Goliath,and then argueby (8) that it is a statue.So we have an endless multiplicationof statues. Secondly, as this last elaborationalso shows, concerningsomethingwhich we originally assumednot to have the modal propertiesof a statue, we have shown that it is a statue, and so does have the modal propertiesof a statue.We started out, as in (2) above, by saying that Lumpl could not survive the loss of those of its parts which make up the arms and calves of Goliath. The groundswere that, unlike Goliath,Lumplis a piece of clay andnot a statue.But by using (8) we have those groundsand in provingthatLumplis a statuewe have thereby undermined establishedthatit could survive the loss of those parts. Surely, if (8) is true,then this is a reductioof the claim thatGoliathis not identical with Lumpl.I0
IV
If (8) is true, then certainrelatedsimilarityprinciplesought also be true. In particular,it ought to be truethat
9 To make such a principle ultimatelydefensible certainrestrictionsnot pertinentto the argumentof the main body of the paperhave to be added.Thus if things intrinsically just like humanbeings grew on trees or sprangfrom the teeth of a certaindragonmany would thinkthat a case could be made for denying them the name of humanbeings since origins of humanbeings. Mutatismutandisfor statues; they do not have the characteristic a statue of Mt. Fujiamathat was full size and made of the same sort of matteras the real Mt. Fujiamamight be intrinsicallyexactly like Mt. Fujiama.But it would be wrong to say thatMt. Fujiamais a statue. (This example is due to Giden Rosen.) For somethingto be a statueit has to be shapedor constitutedwith the intentionof generatingan artobject. But just because it could be said of Lumplthat it was originallyshapedor constituted(by putting togethertwo pieces of clay) with the intentionof generatingan artobject,the required restrictionon (8) would affect the argumentto follow. so thatnothing of "maximality" In a similarvein, some Fs are subjectto a requirement that is a properpartof an F can count as anotherF. The status of such a principleand the requiredqualificationsfor a such a principleare delicate matters.Two tables can make up a thirdtable, and the Pope's crown consists of three crowns. 10 Some might object that the paradigmstatue used to arguethat Lumpl is a statue is none otherthanGoliath,the very statuewhich Lumplconstitutes.But of course a duplicate of Goliath made from the same mold using the same materialwould have done as well.
99
(9) if y is a paradigmF and x is an entity that differs from y in any respect relevantto being an F only very minutely then x is an F. For how can such very minute differences make for the difference between a F and somethingnot an F at all? This principlecan be given strongtheparadigm oretical support in those cases in which "F" is a sortal that can be taught by ostendingparadigms.In such cases, (9) seems to state the groundsfor counting as Fs things other than the ostended paradigmFs. (Imagine that the minute differences do not add up to a perceptibledifference, and that we can recognize Fs by perception.) That (9) is very surprisinglynot true I take to be the upshot of Peter Unger's of the many".The problembegins with the observationthat (1981) nice "problem (10) In the closest vicinity of any paradigmmiddle sized materialF thereare usually very many entities thatdiffer only very minutelyfrom the paradigm in any respect. This is a familiarconsequenceof materialatomism.Unger illustratesthe pointby means of the example of a cloud. Wheneverthere is a cloud or dense cluster of waterdroplets,there is in its closest vicinity a relatively large numberof cloudshaped,equally dense clusters of waterdroplets.All of these clusters are highly coincident,many differing in compositionfrom the others and from the original cloud only in respectof a single droplet.Hence in the closest vicinity of any paradigm cloud there are very many entities-cloud-shaped clusters of water droplets-that differ only very minutely from the paradigmcloud. So, by (9), in the closest vicinity of any paradigmcloud there are very many clouds, in fact an enormousnumber.They are highly coincident, almost completely overlapping, differing only minutely in constitution. Moreover, clouds are here functioning only as a vivid example. The same holds for tables, chairs, trees and animalsanythingthat is materiallycomplex. Rather than conclude that there are enormously many more tables, chairs, trees, etc. than we would ever have dreamtof, Unger himself prefersto say that there are no tables, chairs, trees, etc.-no materiallycomplex Fs for any sort F. So far as I can reconstructthe reasoning,Unger's thoughtis that since termslike etc. are supposedto both function as count nouns, and give rise "table","chair", to reasonablecounts of the numbersof these things, and be governedby principles like (9), they turnout to be not coherentlyusable,thanksto the truthof (10). I think that we can do better than Unger's Eliminative Nihilism. We can explain why (9), thoughplausible,is nonethelessfalse. The explanationwill simfor the main argument ilarly redoundto the discreditof (8), therebyundermining probon Unger's better doing Indeed, identity. takinginvariantconstitutionto be lem will involve us in supposingthatconstitutionis never identity. First we must locate the role played by the phenomenonof vagueness in the problemof the many, lest some thinkthatthe problemis simply and completely a problemof vagueness and will disappearonce one or otherof the more or less adequatetreatmentsof vagueness is applied. It is true that we were adopting a and defpretenseof precisionwhen we supposedthattherewould be a particular
100 MarkJohnston
inite dense cluster of water dropletsconstitutingany paradigmcloud, a definite cluster of which we could say that many other, almost exactly overlapping, cloud-shapedclusters differed from it only minutely in constitution,e.g. by not including one water droplet included in the paradigmcloud. However, the pretense was harmlessfor the purposesof generatingthe problemof the many.The problemof the many still looms if we admitthatour paradigmcloud has no privileged cluster of water droplets as the cluster which exactly constitutes it, the truthbeing insteadthatwith respectto varioussharpenings, or legitimateways of drawingthe boundariesof the cloud, slightly differentsuch clusters will constitute the cloud. For after all, on any sharpening,or way of fixing on a sharply defined cluster to constitute our paradigmcloud, we will have the cloud c, the precise clusterkoand a very large numberof precise clusterskl, k2differingfrom koonly very minutelyand hence differingfrom c only very minutely.Now (9), if true,is definitely true;i.e. it ought to hold on any legitimatesharpening.But then on any legitimate sharpening,such as the one just described, we will have our paradigm cloud c and, by (9), a very large number of almost exactly similar clouds k1,k2... So the problemof the many arises on each legitimatesharpening. What then is the way out? Sticking to the harmlessfantasy of precision, considerthe cloud c andthe clustersko,kl, k2... On any ordinaryway of talking,clusters like ko,kl, k2,etc. are not themselves clouds but may constituteclouds. On ordinaryways of talking,when countingthe numberof clouds we do not contemplate a count of all the distinct, precise, cloud-shapedclusters of water droplets in the nearestvicinity of any cloud. These do not count as clouds. Despite their being quantities of matter which constitute clouds, we do not count them as clouds. Thatis, ourordinaryuse of a count noun like 'cloud' does not strictlyconform to (9) but ratherto (9') If y is a paradigmF and x is an entity that differs from y in any respect relevantto being an F only very minutelyand x is of the right category, i.e. is not a mere quantityor piece of matter,thenx is an F. Hence Unger's insistent and ironic question "But which of ko,kl, k2... is our paradigmcloud c?" has as its proper answer "None". None of these count as clouds because they are mere quantitiesof matterand hence mere constitutorsof clouds. The problemof the many simply shows that constitutionis a vague relation. Our cloud c is not only not identical with any one of ko,kl, k2but also it is not definitely constituted by any one of ko,kl, k2... Rather, on one legitimate sharpeningit is constitutedby one of the ks, on another,anotherof the ks, and so on. What is importantfor our purposes is that on no legitimate sharpeningis c identicalwith any one of the ks. For if thatwere so therewould be a precise cluster k, which on one legitimatesharpeningwas a paradigmcloud and therewould be entities of the same category as kl-all the other precise clusters-which
Constitutionis Not Identity 101
deserve the name of a cloud if k1does. We would be back with the problemof the many. So we see that recognizing that constitution is not ever identity, which involves recognizing a distinctionof categorybetween a materialobject and the matterwhich constitutesit, is crucial in dealing with the problemof the many.'' So also, the alleged reductioof ?3 only serves to show thatinsteadof (8) we need (8'). (8') If y is a paradigmF and x is intrinsicallyexactly like y and x is of the right category, i.e. x is not a mere quantityor piece of matter,then x is an F. (8') reflects the fact thatin countingFs we never dreamof countingas still more Fs the enormousnumberof F-shapedbundlesof matterwhich (on some or other legitimate sharpening)constituteFs. Now perhaps someone with a feel for the problem of the many will admit this as a descriptive claim, but will insist "By what right do we not count such F-shaped bundles of matter as Fs? By what right do we ignore the strikingintrinsicsimilaritiesbetween Fs and their constituting matter,countingFs only on one side of this so-called distinctionof cateI I In understanding the issue as whetherconstitutionis ever identity I have in mind the temporalpart theoristwho admits that in the second case where the head and calves of Goliath are replaced,Goliath is not identical with Lumpl. RatherGoliath and Lumpl, understoodas four dimensional continuants,overlap. Constitutionis the overlap of the temporalparts of persisting quantitiesof matterand of persistingmaterialobjects. Identity is just the case of complete overlap. For a development of this view see Robinson (1982a) and (1982b). Someone who construesidentityas the limiting case of this kind of overlap will not be able to justify any distinctionof ontic type between a materialobject and its constitutingmatter.For on this view it will sometimes be true that a materialobject's constitutingmatteris identicalwith thatmaterialobject. Since constitutingmatteris is all of the same category and if x=y then y is of the same category as x, it follows that some material objects are of the same category as pieces of constituting matter.So the problemof the many looms. The version of the problem of the many which holds most directly against the view verthat identity is the limiting case of overlap of temporalpartsis the four-dimensional sion. Think of a four-dimensionalsum of cloud stages that is supposed to be a paradigm persisting cloud. In the paradigm'snearest spatiotemporalvicinity there are very many sums of cloud stages which differ only very minutelyfrom the paradigm.The problemof the many is underwayif on a legitimate sharpeningthe persistingcloud is identical with one of the sums, so that one of these sums deserves the name of a persistingcloud. Why don't the others? What of the idea (suggested by David Lewis) that they do, but that this is harmless since, because of massive overlap,the many are almost one? Well, I do not thinkthatthis gives as general solution. Michael's house has an extension almost as large as his original house, which in fact was a nearduplicateof the next door neighbour'shouse. While some have houses consisting of two houses (a haciendaoften consists of the main house and the dwelling made larger servants'gatehouse)Michael doesn't. His house is a single suburban than it originally was by an extension. Yet (8) implies that Michael has (at least) two houses-the large house with the extension, and the propersubpartof that house which made up the originalhouse and is intrinsicallyjust like the next door neighbour'shouse. The point is that these two houses do not massively overlap, they are not almost one. Michael's extension is almost as large as his originalhouse. I conclude thatthe "manybut almost one" solution does not work here. (8) needs restrictinganyway, and when the right restrictionis made the originalproblemof the many is blocked along with the problemof Michael's house.
102 MarkJohnston
gory? What is there to the distinction anyway once we admit that there is nothing more to an F than its constituting matter?Isn't this just a distinction without a difference?"
V Whatgives these questionsforce is theirappealto a particular conceptionof what is requiredto justify certain systematicpatternsand demarcationsin our judgeThe insistenceon no disments of identity,differenceand sortal-relative number. tinction between an F and its F-shaped constitutingmatterwithout a difference to justify the distinction is uncontroversialon its face. It is a certain seductive conceptionof what a justifying differencemust consist in which ought to be controversial.The conceptionhas it thatthejustifying differencehas to be substantial and characterizable independentlyof our practiceof makingjudgements which exhibit certainpatternsand demarcations.So on this substantialconception it is not enoughfirst to say thatthe difference between an F and its F-shaped constituting matter is just the difference we mark when for example we standardly don't countF-shaped constitutingmatteras an F, and then having said thatgo on to give an internaland pragmaticjustificationof this in terms of how a practice which marksthis distinctionserves ourpurposes.In contrastwith such a minimal and practice-dependentjustification, practice-dependent because it only has recourseto ourpracticesandtheirinternalrationales,the substantial andindependent justification would resort to some metaphysical model of the relation between an F and its F-shapedconstitutingmatter.The practice-independent justification of the distinction will then be that there is some metaphysical extra which an F includes and an F-shaped bundle of matterdoes not. The catalogue of proposed extra ingredientsis none too impressive. Among the entities proposed as ingredientsin Fs along with their matter,we have had bare, that is non-qualitative, but somehow essentially F-ish particulars,F-ish substratawhich hold togetherthe properfeaturesof Fs as if those featureswere like pins in a pin cushion, and haecceities or particular identity properties,such as being Mt. Fujiama. Not thatall of these ideas arejust silly. Most of them have their origins in reasonablethoughts about materialobjects which then become perverted by the demandfor a metaphysicalmodel of a materialobject which discerns extra ingredientsbesides the object's constitutingmatter.So, for example, bareparticulars or haecceities as ingredientsis a misthe idea of non-qualitative taken hypostasizing of a rathergood idea, which has of late been called (along with many other different ideas of varying merit) haecceitism,'2 the idea that
12 For various versions of haecceitism see Kaplan (1975), Adams (1979) and Lewis (1986, ?4.4.).
there is no purely qualitativecriterionfor the identityof objects. So too the idea thateach F includes as an ingredientbesides its constitutingmattera substantial form, which unlike the F's constitutingmatteris essentially F-ish, is a mistaken hypostasisof the rathergood idea thatif we have objects classified into substance sorts or relatively naturalkinds and F is one such sort or kind then we will not countenanceas possible a situationin which somethingthatis actuallyanF exists ways. Justas we can have haecwithoutbeing in thatsituationF-ish in important or haeccieties as extra ingredients,we can have ceitism withoutbare particulars essentialism withouthaving substantialforms or essences as extraingredients. Alternativeconclusions can be drawnfrom surveying at once both the catalogue of proposed metaphysical extras and the problem of the many. The first conclusion is thatthereis no crediblemetaphysicalextraand thereforeno justifiable distinction to be made between an F and its constitutingmatter.Then the problem of the many is a straightproof to the effect that our practice of using count nounsor sortalsis incoherent.The betterconclusion is thatourpracticeand the distinctionit embodies is acceptableas it standsand what is bogus is the conception of justifying our practicewhich requiresthat,for the distinctionto be justified, the difference between an F and its constituting matter must be a deep metaphysicaldifference securedby an extra ingredientof the F. drawsthe second conclusion. Here, It is the Minimalistwho characteristically as elsewhere, he aims for ontology withoutmetaphysics,which is to say general talk aboutrealitywithoutthe postulationof extraingredientswhich it is the peculiar privilege of philosophy to discover.The Minimalistaccuses both the Eliminativist like Unger and the Hyperrealist,who believes in extra ingredients, of making the same error,the errorof supposingthatour practiceof distinguishing Fs and theirconstitutingmatterand countingaccordinglycould only be justified 3 if the distinctionis securedby the independentmetaphysicsof the matter.' What then does justify our practiceof employing the distinctionand counting the world as made up of material accordingly?Probablyonly this: in representing objects of various sorts,persistingthroughall but substantialchanges in respects crucial to their sort, we provide for ourselves certainexplanatorystrategies.We are able to explain sortal relative continuities in terms of the persistence of objects of the relevantsorts, and to explain change and discontinuityin terms of the naturesand interactionsof persisting objects of various sorts. But since the middle-sized persisting objects of our experience are materially complex and change constantlyundergomaterialchange which falls shortof being substantial in respects important to objects of their sort, we cannottake them to be identical with the parcels or bundles of matterwhich make them up at various times. For often, as with the case of Goliathand Lumpl, such materialchanges are substantial changes in respects importantto the persistence of bundles or quantitiesof
13 For more on the advantagesof Minimalism see Johnston(1987). Unger seems to have become more sympatheticto Minimalismin his (1990) in which Minimalistthemes are deployed against Derek Parfit'sattemptto derive revisionaryconsequences in ethics from the observationthat we are not "seperatelyexisting entities distinct from our brains and bodies".
104 Mark Johnston matterin question.Goliathis thus typical in being able to survivechanges which the piece of matterwhich originally constitutedit cannot survive. Hence within our scheme of persistingobjects and theirconstituentpieces or quantitiesof matter thereare systematicreasonsto distinguishobjects from the filatterwhich constitutesthem. The crucial distinctionis a consequence of representingthe world as a worldof bothobjects andpieces of mattersurvivingmaterialchange,but surviving differentkinds and degrees of materialchange. The explanatorystrategieswhich this style of representation makes possible are sometimescontrastedwith an explanatorystrategywhich accountsfor change -andcontinuityin terms of the laws governing the propagationof materialproperty instancesover space-time.This last style of explanationhas been extraordinarily successful in serving the purposes of modern science. But there is an enormousgulf, so far unabridged by argument, falling between properlyrespecting that style of explanation and accepting scientism, which in this context amountsto the view thatthere are only pattems of distributionof materialproperty instances,so thatpersistingobjects, if they are to exist at all, must be identical with some such pattems. In the absence of a defense of scientism, we can rest content with our scheme of persistingobjects distinctfrom theirconstitutingmatter.After all, the scheme has served us very well in making intelligible a world of change and continuity. Russell said that we can properlydo without the metaphysics of the stone age. Just so, but we can also properlydo without the scientistic metaphysics of our own age. The ontology of the stone age is however anothermatter.It, fortunately, is still with us and well deserves its place.'4 Departmentof Philosophy 1879 Hall Princeton University Princeton NJ 08544-1006 USA REFERENCES Journal of PhiAdams, R. M. 1979: "PrimitiveThisness and PrimitiveIdentity". losophy, 76. pp. 5-26, Gibbard,A. 1976: "ContingentIdentity".Journal of Philosophical Logic, 4. pp. 187-221 Gupta,A. 1980: TheLogic of CommonNouns. Yale, NH: Yale UniversityPress.
'4 I wish to thank Ned Hall, Saul Kripke, David Lewis, Denis Robinson, Gideon Rosen, SarahStroud,Jamie Tappenedenand Peter Unger for useful discussions on these topics.
MARK JOHNSTON
Johnston,M. 1987: "Is There a ProblemAbout Persistence?" Proceedings of the AristotelianSociety. Supp. 61 pp. 107-135 Kaplan,D. 1976: "How to Russell a Frege-Church". Journal of Philosophy, 72, pp. 716-729 1. Leonard, H. S. and Goodman, N. 1940: "The Calculus of Individuals and Its Uses". TheJournal of SymbolicLogic, 5, pp 46-7. Lewis, D. K. 1968: "Counterpart Theory and QuantifiedModal Logic", Journal of Philosophy, 65, pp. 113-126 1986: On the Plurality of Worlds.London:Basil Blackwell. 1971: "Counterparts of Persons and Their Bodies". Journal of Philosophy, 68, pp. 203-211 Robinson, D. 1982a: The Metaphysicsof Material Constitution.Ph.D. Dissertation, MonashUniversity. 1982b:"Re-IdentifyingMatter". Philosophical Review, 81, pp. 317-342 Thomson, J. J. 1983: "Parthood and IdentityAcross Time". TheJournal of Philosophy, 80, pp. 201-219 Unger, P. 1981: "The Problemof the Many"in MidwestStudies in Philosophy, Vol. VI, French,P. A., Uehling, Jr.T. E., and Wettstein,H. K., eds., University of MinnesotaPress, Minneapolis. 1990: Persons, Consciousnessand Value.Oxford:OxfordUniversityPress. Williams, D. C. 1966: "TheElementsof Being" in Principles of EmpiricalRealism, Ruja,A., ed., Springfield:CharlesThomas. 1986: "Universalsand Existents".AustralasianJournal of Philosophy, 64, pp. 1-14

Constitution Is Not Identity

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Constitution Is Not Identity

Diunggah oleh

Hak Cipta:

Format Tersedia

Mind Association

ConstitutionIs Not Identity

Mind, Vol. 101 . 401. January 1992

Constitutionis Not Identity

Constitutionis Not Identity

and So it seems thatif we want to give "O(x,y)"its properlogical interpretation, by means of

O(z,y)) is logically equivalent to x=y.

Constitutionis Not Identity

Constitutionis Not Identity

Constitutionis Not Identity 101

Constitutionis Not Identity 103

Constitutionis Not Identity 105

Anda mungkin juga menyukai