Journal of Philosophy, Inc.

Counterparts of Persons and Their Bodies Author(s): David Lewis Reviewed work(s): Source: The Journal of Philosophy, Vol. 68, No. 7 (Apr. 8, 1971), pp. 203-211 Published by: Journal of Philosophy, Inc. Stable URL: http://www.jstor.org/stable/2024902 . Accessed: 19/08/2012 18:48

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 [email protected].

.

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Philosophy.

http://www.jstor.org COUNTERPARTS OF PERSONS AND THEIR BODIES 203

COUNTERPARTS OF PERSONS AND THEIR BODIES * t M/rATERIALISTS like myselfhold that personsand their bodies are identical. But there is a simple argumentto show that thisidentity thesis is refutedby the merepossi- bilitythat a person mightswitch bodies. To defeat the argumentit seems necessaryto revise my counterparttheory by providingfor a multiplicityof counterpartrelations. This revisionhas an odd result. Modal predictionsmay be de re, yet not referentiallytransparent. The thesisI wish to defendhere may be stated moreprecisely, as follows: (T) Necessarily,a personoccupies a bodyat a timeif and onlyif that personis identicalwith that body at that time. Note that the thesis (T) is formulatednot in termsof identityitself, a two-placerelation, but in termsof a derivativethree-place relation of identityat a time.I wish to regardenduring things such as persons and bodies as aggregates-sets, mereologicalsums, or something similar-of momentary stages. Enduring things X and Y are identicalat a timet ifand only if theyboth have stages at t-that is, existat t-and theirstages at t are identical.Therefore X and Y are identicalsimpliciter if and only if theyare identicalwhenever either one exists. Note that (T) does not say that personsand bodies must be identical simpliciter.It does imply that if a certain person occupies a certain body whenevereither the person or the body exists,then the personand the body are identical.In such a case, all and only thosestages whichare stages of the personare stages of the body he occupies. But (T) also permitsother cases: forinstance, a body consistingof the stages of a certainperson togetherwith some finaldead stages that are not stages of any person (and some initial prenatal stages that perhaps are not stages of any person); or a person consistingof stages of a certain body togetherwith some initialor finalghostly stages that are not stages of any body; or even a body-switchingperson consisting partly of stages of one body and partlyof stages of anotherbody. A person consistsof stages related pairwise by a certain relation we may call the relationof personal unity;a body consistsof stages related pairwiseby anotherrelation we may call the relation of bodilyunity.' Since for the most part * I amindebted to M. J.Cresswell, David Kaplan, and John Perry for illuminat- ingdiscussion of these matters. 1 Theso-called "problem of personal identity" is theproblem of explicating the relationof personal unity between stages. This view is expoundedmore fully by John Perry in "Can the Self Divide?", presented at the APA Pacific Division meetings, Mlarch1970. 204 THE JOURNAL OF PHILOSOPHY persons occupy bodies and bodies are occupied by persons, it followsaccording to (T) that the two relationsof unityare relations on almost the same set of stages. The exceptionsare dead stages, perhapsprenatal stages, and perhapsghostly stages. Moreover,if we leave out thedead or ghostlyor perhapsprenatal stages, then at least for the most part the two relationsof unity are coextensive.The exceptionswould be body-switchersand perhaps split personalities. Nevertheless,the two relations of unity are differentrelations-in- intension;so theyare coextensiveonly contingentlyif at all. Now I shall present an argumentagainst (T). I regard it as a simplifieddescendant of an argumentput forthby JeromeShaffer,2 but I have changed it so much that he mightnot wish to acknowl- edge it as his own. Body-switchingis logically possible. Because I might have switchedout of my presentbody yesterday,though in fact I did not, I and my actual presentbody are such that the formermight not have occupied the lattertoday. Whetheror not personsare identical with bodies they occupy, certainlypersons are never identicalwith bodies they do not occupy. So we have: (1) I and my body are such that theymight not have been identical today. Suppose that, as is surelyat least possible, I occupy the same body fromthe timewhen it and I began until the timewhen it and I will end. Then, by (T), it and I are identicalwhenever it or I exist.Hence my body and I, enduringthings, are identical simpliciter.By this identityand Leibniz's law, (1) yields (2): (2) My body and my body are such that theymight not have been identicaltoday. Since (2) is self-contradictory,(T) has apparentlybeen refuted. To rescue (T) withoutdenying the possibilityof body-switching and withoutdenying that I mightoccupy the same body throughout the time that it or I exist, I plan to show that the step from(1) to (2) by Leibniz's law is illegitimate-in otherwords, that (1) is not referentiallytransparent with respectto the term'I'. I have used the familiar"are such that" constructionto indicate that (1) and (2) are to be taken as de re ratherthan de dictomodal predications.That is, we are to considerwhat happens in otherpos- sible worlds to the thingsdenoted herein our actual world by the terms 'IT and 'my body', not what happens in other worlds to the 2 "Personsand TheirBodies," PhilosophicalReview, LXXV, 1 (January1966): 59-77. COUNTERPARTS OF PERSONS AND THEIR BODIES 205 things denoted thereby those terms. Suppose (1) were taken de dicto,as if it were this: It mighthave been the case that I and mybody were not identical today. There is no problemexplaining why this is not referentiallytrans- parent. Its truthconditions involve the denotationsin otherworlds of the terms'I' and 'my body'; even if these denote the same thing herein our world,they denote differentthings in some otherworlds, forinstance in a worldin whichI switchedbodies yesterday.Hence they are not interchangeable.But it would be wrong to take (1) as de dicto,for the argumentleading up to (1) would then be in- coherentand question-begging.As we understand (1) in the argu- ment,it seems true even given (T) because of the fact that I might have switchedbodies yesterday.But (1) taken de dictois a straight- forwarddenial of an instance of (T), and the possibilityof body- switchingis irrelevantto its truthor falsity.For (1) taken de dicto is true today if and only if thereis some worldsuch that the things denoted in that world today by 'I' and 'my body' are not identical today. But in any world,'my body' today denotes whicheverbody is today occupied by the personwho is today denotedby 'I', regard- less of whetherthat person occupied that body yesterday.If (T) is true, then in any world the person today denoted thereby 'I' and the body today denoted thereby 'my body' are identical today; so (1) taken de dicto is false. (It should be understoodthat when I speak of the denotationin anotherworld of 'IT and 'my body', I am not concernedwith any utteranceof these termsby some inhabitant of the otherworld, but ratherwith the denotationof these termsin the otherworld on the occasion of theirutterance by me here in our world.) I have suggestedelsewhere3 that de re modal predicationsmay best be understoodby the methodof counterparts. To say that something here in our actual world is such that it mighthave done so-and-so is not to say that thereis a possible worldin which that thingitself does so-and-so,but that there is a world in which a counterpartof that thingdoes so-and-so.To say that I am such that I mighthave been a Republican, but I am not such that I might have been a cockatrice,is to say that in some world I have a counterpartwho is a Republican, but in no world do I have a counterpartwho is a cockatrice.That is plausible enough,for the counterpartrelation is a relationof similarity.X's counterpartsin otherworlds are all and 3 "CounterpartTheory and QuantifiedModal Logic," this JOURNAL, LXV, 5 (March 7, 1968): 113-126. 206 THE JOURNAL OF PHILOSOPHY only those things which resemble X closely enough in important respects,and more closely than do the other thingsin theirworlds. It is easier for a Republican than for a cockatriceto resembleme enough to be my counterpart. The counterpartrelation serves as a substitutefor identitybe- tween things in differentworlds. The principal advantage of the method of counterpartsover the method of interworldidentities is that if we adopted the latter in its most plausible form,we would say that thingswere identicalwith all and only those thingswhich we would otherwisecall theircounterparts. But that could not be correct: first,because the counterpartrelation is not transitiveor symmetric,as identityis; and second, because the counterpartrela- tion depends on the relative importanceswe attach to various differentrespects of similarity and dissimilarity,as identitydoes not. To recapitulate:in each possibleworld there is a set of momentary stages and a set of enduringthings composed of stages relatedpair- wise by various relationsof unity.An enduringthing and its stages exist only in one world,but may have counterpartsin otherworlds. We shall be concerned here only with counterpartsof enduring tlhings,though we can allow that stages also have theircounterparts. Applyingthe methodof counterpartsto the problemat hand, we immediatelyencounter a bothersomedistraction. The translation of (2), whichseemed self-contradictory,is this: There are a world W, a counterpartX in W of my body,and a counterpartY in W ofmy body, such that X and Y are notidentical today. Unfortunately,this translationcomes out true,but foran irrelevant reason. I, and also my body whetheror not I am identical with it, mighthave been twins. My body thereforedoes have two different counterpartsin certain worlds. Not only is the translationtrue; it seems to me to show that (2) itself is true. But the argument against (T) can easily be repaired.Replace (1) and (2) by: (1') I and my bodyare such that (withoutany duplicationof either) theymight not have beenidentical today. (2') My bodyand my bodyare such that (withoutany duplicationof either)they might not have beenidentical today. The argumentworks as well with (1') and (2') as it did with (1) and (2). Indeed, (1') and (2') correspondto (1) and (2) as we would have understoodthem if we had forgottenthat I mighthave been twins.4 4 I do notknow how or whether (1') and (2') can be expressedin thelanguage of quantifiedmodal logic. That does notbother me. I knowhow to expressthem in Englishand in counterparttheory. COUNTERPARTS OF PERSONS AND THEIR BODIES 207

Applyingcounterpart theory to the repairedargument, we obtain these translationsof (1') and (2'): (1*) Thereare a worldW, a uniquecounterpart X in TVof me,and a uniquecounterpart Y in W ofmy body, such that X and Y are not identicaltoday. (2*) Thereare a worldW, a uniquecounterpart X in W ofmy body, and a uniquecounterpart Y in W of mybody, such that X and Y are notidentical today. The argumentagainst (T) seems to go through,using (1*) and (2*): (1*) seems true because I might have switched bodies yesterday; (2*) is self-contradictory;yet (1*) implies (2*) by Leibniz's law, given (T) and the suppositionthat I occupy the same body when- ever I or it exist. In defenseof (T), however,I claim that (1*) is false, despite the fact that I might have switched bodies yesterday.What is true because I mighthave switchedbodies is not (1*) but rather (1**): (1**) Thereare a worldW, a uniquepersonal counterpart X in W of me, and a uniquebodily counterpart Y in W of mybody, such that X and Y are notidentical today. I now propose a revisionof counterparttheory to the effectthat, at least in the presentcontext, (1**) ratherthan (1*) is the correct translationof (1'). What followsfrom (1**) by Leibniz's law, given (T) and the suppositionthat I occupy the same body wheneverI exist,is not the self-contradiction(2*) but ratherthe truth: Thereare a worldTV, a uniquepersonal counterpart X in W of my body,and a uniquebodily counterpart Y in W ofmy body, such that X and Y are notidentical today. Two othertruths follow from (1**) in the same way: Thereare a worldW, a uniquepersonal counterpart X in WVof me, and a uniquebodily counterpart Y in W ofme, such that X and Y are notidentical today. Thereare a worldTV, a uniquepersonal counterpart X in W of my body,and a uniquebodily counterpart Yin W ofme, such that X and Y are notidentical today. However, the translationof (2') is none of these. If the translation of (1') is (1**), the translationof (2') should be (2**): (2**) Thereare a worldW, a uniquebodily counterpart X in W of my body,and a uniquebodily counterpart Y in W of mybody, such thatX and Y are notidentical today. Though (2**) is not (2*), it is still a self-contradiction. 208 THE JOURNAL OF PHILOSOPHY

As we already noted,counterpart relations are a matterof over-all resemblancein a variety of respects. If we vary the relative im- portancesof differentrespects of similarityand dissimilarity,we will get differentcounterpart relations. Two respectsof similarityor dis- similarityamong enduringthings are, first,personhood and personal traits,and, second, bodyhood and bodily traits. If we assign great weightto the former,we get the personalcounterpart relation. Only a person,or somethingvery like a person,can resemblea person in respectof personhoodand personal traitsenough to be his personal counterpart.But if we assign great weightto the latter,we get the bodilycounterpart relation. Only a body, or somethingvery like a body, can resemblea body in respectof bodyhoodand bodily traits enough to be its bodily counterpart. If I am my body, then in many worldsthere are thingsthat are both personal and bodily counterpartsof me and ipso facto of my body. These things,like me, are both persons and bodies. But in other worlds I (and my body) have neitherpersonal counterparts nor bodily counterparts; or personal counterpartsthat are not bodily counterparts;or bodily counterpartsthat are not personal counterparts; or personal and bodily counterpartsthat are not identical.A worldin which I switchedout of my body-that is, my personal counterpartswitched out of my bodily counterpart- yesterdayis of this last sort. I and my body have therea personal counterpartthat is a person but not a body and also a bodily counterpartthat is a body but not a person.These are not identical today, and not identical simpliciter,though they were identical at times beforeyesterday since they shared theirearlier stages. How- ever, my personal counterpartis identical today with a different body. My bodily counterpartis identical today with a different person (if the body-switchingwas a trade) or with none. We may draw an analogy between the relationsof personal and bodily unity among stages of persons and bodies and the personal and bodilycounterpart relations among enduring persons and bodies. If I ask of somethingthat is both a stage of a body and a stage of a body-switchingperson "Was this ever in Borneo?" you should ask whetherI mean this person or this body. If the former,I am asking whetherthe given stage is linked by personal unity to an earlierstage located in Borneo. If the latter,I am askingwhether it is linked by bodily unity to an earlier stage located in Borneo. Similarly,if I ask of somethingthat is both an enduringperson and an enduringbody "Might thishave been an orangutan?"you should again ask whetherI mean this personor this body. If the former,I COUNTERPARTS OF PERSONS AND THEIR BODIES 209

am asking whetherit has an orangutanfor a personal counterpart; if the latter,whether it has an orangutanfor a bodily counterpart. But the analogy is imperfect.The two relations of unity are equivalence relations,at least forthe most part and as a matterof contingentfact. Therefore it is easy and natural to formthe concept of an enduringperson or body, consistingof stages linked together pairwiseby a relationof personalor bodily unity. It is temptingto do the same with the counterpartrelations, forming the concept of a superpersonor superbody consistingof persons or bodies in differentworlds, linked together by a personalor bodilycounterpart relation. But this cannot be done in any straightforwardway be- cause counterpartrelations are not equivalence relations. Like all similarityrelations on a sufficientlyvariegated domain, they fail to be transitive because chains of little differencesadd up to big differences. Why should I thinkit plausible to employ multiplecounterpart relations to translate (1') as (1**) rather than (1*)? Precisely be- cause by doing so I escape the refutationof (T), and I am convinced of (T). I am offereda trade: instead of a multiplicityof kinds of thingI can have a multiplicityof counterpartrelations. A reductio refutesthe whole combinationof assumptionsthat led to contradic- tion; if all but one of those assumptionsare highlyplausible, which- ever remainsis the refutedone. And in addition,if I contemplatethe propositionsI express by means of (1') and (1**), it seems to me that they are the same. I would like to presentthe translationof (1') by (1**) and (2') by (2**) as instancesof a generalscheme for translating English modal predications into sentences of counterpart theory with multiple counterpartrelations. I do not know how to do this. Roughly, the idea is that the sense of a termsomehow selects the counterpartrela- tion that is to be used to findthe counterpartsof the thingdenoted by that term.The terms'I', 'you', 'that person','the lady I saw you withlast night','George', all select the personalcounterpart relation. 'This thing' (pointingat myself),'this body', 'my body', 'that which will be my corpse afterI die', all select the bodily counterpartrela- tion. Similarlyfor indefinite terms (phrases of restrictedquantifica- tion): 'everybody'selects the personalcounterpart relation, whereas 'every body' selects the bodily counterpartrelation. Even if every- body is his body, and conversely,"Everybody is such that he might have been a disembodiedspirit" is true,whereas "Every body is such that it mighthave been a disembodiedspirit" is false. The former means that each of those thingswhich are both personsand bodies 210 THE JOURNAL OF PHILOSOPHY has a disembodiedspirit as personalcounterpart, whereas the latter means that each of those same thingshas a disembodiedspirit as bodily counterpart. In certain modal predications,the appropriatecounterpart rela- tion is selected not by the subject termbut by a special clause. To say that something,regarded as a such-and-such,is such that it mighthave done so-and-sois to say that in some worldit has a such- and-such-counterpartthat does so-and-so.With these "regardedas" clauses in mind, I mightsay that I translate (1') as (1**) because I take it to be synonymouswith (1"): (1") I, regardedas a person,and mybody, regarded as a body,are such that (withoutany duplicationof either) they might not have been identicaltoday. Likewise I translated (2') as if it had contained two "regarded as a body" clauses. If we are to have multiple counterpartrelations, we may well wonder how many to have. One for every sortal? One for every natural kind? One forany commonnoun phrase whateverthat can grammaticallybe insertedinto 'regardedas a ', even the phrase 'yellowpig or primenumber'? One forany kindof entity,even kinds that cannot be specifiedin our language?5I do not know. Nor do I know whetherone of the counterpartrelations, corresponding per- haps to the clause 'regardedas an entity',can be identifiedwith the single counterpartrelation of my originalcounterpart theory. It is customaryto distinguishreal essences of thingsfrom their nominal essences under descriptions.Now, however, we have a third,intermediate, kind of essence. My real essence consistsof the propertiescommon to all my counterparts.(Here I use the original singlecounterpart relation.) My nominalessence under the descrip- tion 'person' consists of the properties common to all possible persons. My intermediateessence under the description'person' consistsof the propertiescommon to all my personal counterparts. I have no reason to thinkthat any two of these sets of properties are the same. It may even be that none of the threeis properlyin- cluded in any other, if my personal counterpartsinclude some entities (robots, say) which are almost persons but not quite. Counterpartrelations are vague, being dependent on the relative weightsassigned to respectsof similarityor dissimilarity.Hence real 6 We couldalso putthis question another way: givena three-placerelation if is a al counterpartof..", whichkinds are appropriatemiddle arguments? COUNTERPARTS OF PERSONS AND THEIR BODIES 2 I essences are vague in a way nominalessences are not. Intermediate essences under descriptions share this vagueness, for the new multiplecounterpart relations are no less vague than the original counterpartrelation. In my originalcounterpart theory, any de re modal predicationis referentiallytransparent. Something has the same counterpartshow- ever we may choose to referto it. Given a de re modal predication, we findthe thingdenoted by the subject termin the actual world; then we consider what befalls that thing-or rather,its counter- parts-in other worlds. Only the denotation of the subject term matters. We can substitute another subject term with the same denotation but differentsense, and the truth value of the modal predicationwill not change. But in the present revision of counterparttheory, de re modal predicationsare not in generaltransparent. Not only the denotation of the subject term matters,but also the counterpartrelation it selects. If we substituteanother subject termwith the same denota- tion but differentsense, it may change the truthvalue of the modal predicationby selectinga differentcounterpart relation. Then even though the denoted thing here in our world remains the same, we have a differentway of followingthe fortunesof that thingin other worlds. Nevertheless,these modal predicationsare still de re,not de dicto. We still find the denoted thing in our actual world and then find counterpartsof that thingelsewhere. We do not at all considerthe thingsdenoted by the subject termin other worlds,as we would in the case of a de dictomodal predication. Transparencyof modal predicationscan fail wheneverthe sense of the subject termis used to do anythingbeyond determiningthe actual denotationof the subject term.One furtherthing it mightdo is determinethe denotationof the subject termin otherworlds; that is the de dictocase. Another,and altogetherdifferent, further thing it mightdo is select a counterpartrelation. (These two are not the only alternatives.) It is the latter, I suggest, that happens in the argumentwe are considering.Therefore we can accept (1') as a consequence of the possibilitythat I mighthave switched bodies, reject (2') as self-contradictory,and yet accept (T) and its conse- quence that if I occupy the same body wheneverI or it exist then I am my body. DAVID LEWIS St. Catherine's College, Oxford