Вы находитесь на странице: 1из 12

The Calculus of Individuals and Its Uses Author(s): Henry S.

Leonard and Nelson Goodman Source: The Journal of Symbolic Logic, Vol. 5, No. 2 (Jun., 1940), pp. 45-55 Published by: Association for Symbolic Logic Stable URL: http://www.jstor.org/stable/2266169 Accessed: 08/07/2010 15:22
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=asl. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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.

Association for Symbolic Logic is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Symbolic Logic.

http://www.jstor.org

LOGIC or Tax JoURNAL Saouc Volume 5. Number 2, June 190

THE CALCULUS OF INDIVIDUALS AND ITS USES1


HENRY S. LEONARD AND NELSON GOODMAN

in I. An individualor wholewe understandto be whateveris represented any given discourse by signs belongingto the lowest logical type of which that discoursemakes use. What is conceivedas an individualand what as a class is thus relativeto the discoursewithinwhichthe conceptionoccurs. One task of which entitiesare to be construedas individuals applied logic is to determine and whichas classes when the purpose is the developmentof a comprehensive systematicdiscourse. The conceptof an individualand that of a class may be regardedas different one all from thatremains. of devicesfordistinguishing segment thetotal universe divisible,and may even segmentis potentially In both cases, the differentiated in be physicallydiscontinuous. The difference the concepts lies in this: that no to conceive a segmentas a whole or individualoffers suggestionas to what these subdivisions,if any, must be, whereas to conceive a segmentas a class schemeof subdivision-into subclasses and members.2 imposesa definite The relationsof segmentsof the universeare treated in traditionallogistic at two places, firstin its theoremsconcerningthe identityand diversityof and class-inclusion. But individuals,and second in its calculus of membership demand consideration. relationsof segmentsand of classes frequently further For example,whatis therelationofthe class ofwindowsto the class ofbuildings? of No member eitherclass is a memberof the other,nor are any of the segments isolated by the other. Yet isolated by the one conceptidenticalwithsegments relationin that each window is a the classes themselveshave a very definite part of some building. We cannot expressthisfactin the language of a logistic relationbetweenindividualsunless,by makinguse of whichlacks a part-whole we some special physicaltheory, raise the logical type of each windowand each buildingto the level of a class-say a class of atoms-such that any class of in atoms that is a windowwill be included (class-inclusion) some class that is a upon the dependence of logical formulation building. Such an unfortunate the presumpor discoveryand adoption of a special physicaltheory, even upon could in everycase be discoveredin the courseof tionthat such a suitabletheory a time, indicates serious deficienciesin the ordinarylogistic. Furthermore, raisingof type like that illustratedabove is oftenprecludedin a constructional ideas. the governing choice of primitive systemby otherconsiderations
Received July 28, 1939. 1 A somewhat elaborated version of a paper read in Cambridge, Mass., before a joint meeting of the Association forSymbolic Logic and the American Philosophical Association, Eastern Division, on December 28, 1936. 2 The relation is somewhat analogous to the more familiar one between classial and serial concepts, dealing as they do with the same material, but in a manner that makes the latter more highly specialized. 45

46

HENRY

S. LEONARD

AND NELSON

GOODMAN

The ordinary no logisticdefines relationsbetweenindividualsexceptidentity and diversity. A calculus of individualsthat introduces otherrelations, such as the part-whole relation,would obviouslybe veryconvenient;'but what chiefly concernsus in this paper is the generalapplicabilityof such a calculus tothe solutionof certainlogico-philosophical problems. The calculus of individualswe shall employis formally from indistinguishable the general theoryof manifoldsdeveloped by Lesniewski.4 Leiniewski's purfromours, was to establisha generaltheoryof manifolds pose, quite different that would not be subject to Russell's paradox; but sincehe excludesthenotion of a null class, his formal systemis virtually same as thatwhichwe interpret the as a calculus of individuals. Inasmuch as his system is ratherinaccessible, lacks many usefuldefinitions, is set forth the language of an unfamiliar and in logical doctrine and in wordsratherthan symbols, shall attempt(in Part II) we to restate the calculus in more useable form,with additional definitions, a practicalnotationand a transparent Englishterminology. In Part III we shall explainhow this calculus enables us to describegenerally certainimportant, but of oftenneglectedproperties relations,and thereby contributes the clarificato tionofmanyphilosophical problems. of II. The generalfeatures the abstractcalculus may perhapsbe mostreadily apprehendedby comparisonwith the Boolean algebra of classes. It involves operations of addition, multiplication,and negation, a part-wholerelation and an elementanalogous to the Boolean universal analogous to class-inclusion from Boolean analogue in ways consequentupon the refusal the class. It differs to postulate a null element,althoughthe primitive relationof "discreteness" may be correlatedwith the Boolean function"x*y= 0". In the lightof thisanalogy,the characteristic propositions the calculusmay of described: To any analyticproposition the Boolean algebra will be generally of a correspond postulate or theoremof this calculus providedthat, when in the Boolean propositionevery expressionof the form"x y=0" is replaced by an of expression the form"x is discretefromy", no reference the null element to remainsand everyproductand negationis eitherdeduciblyunequal to the null to elementor else is conditionally affirmed be unequal to it. From the three postulates(presented below) of the formal calculus,enoughtheorems have been deduced to indicate that this characterization accurate. Some illustrative is theorems appear in thesequel. of For the formalestablishment the calculus, the symbolism and logisticof Whiteheadand Russell's Principia mathematics have been employedin orderto secure correlationwith other logical doctrines. Only the one primitiveidea already mentionedis required: the dyadic propositional or function, relation, written to "xly" and hereinterpreted mean that the individualswhichare its have no part in common, that theyare discrete.' In our interpretaarguments
3Since this paper was presented, the convenience of such a calculus of individuals has been well illustrated by Dr. J. H. Woodger's Axiomatic method in biology (1937). 4 In 0 podstawach matematyki (in Polish), Przeglqd filozoficzny,vols. 30-34 (1927-31). 6 Legniewski employs discreteness as his primitive relation in the final version of his system. See Chapter X of his above-mentioned paper.

CALCULUS OF INDIVIDUALS

AND ITS USES

47

be partsand commonpartsneed not necessarily spatial parts. tion,furthermore, two concrete problems, Thus in our applicationsof the calculus to philosophic entities,to be taken as discrete,have not only to be spatially discrete,but also discrete,discretein color,etc., etc. temporally In terms of the one primitiveidea just described,other concepts may be definedas follows: I.01
X < Y =Df .Z l y DFZ l

I.e., one thingis a part of anotherif whateveris discretefromthe latteris also discretefromthe former. I.011 I.02
? X << Y Df X < y .

X id Y

partin a familiarsense. proper This defines


X O Y =Df (3z) . z < x .z < y

I.e., two things overlap if they have a part in common. As postulate I.13 dis(below) indicates,overlappingis equivalent to the denial of the primitive creteness. I.03 x Fu
a =Df Z lX y ea

D, Z

relationof an individual,x, whichis the fusion, This definesthe heterogeneous of or sum-individual, a class, a, to that class. An individualis said to stand in that is discretefromit is discretefrom that relationto a class wheneverything discretefromevery memberof the every memberof the class and everything it. Postulate I.1 affirms that any not null class has a sum, class is discretefrom is that the relationheredefined a in and it is demonstrated subsequenttheorems of one-manyrelation,so that we may speak of thesum, or fusion, any existent class, a. I.04 x Nu a
=Df

Z <

X .

/ EY a

Dy Z < Y

This definesthe relation of an individual which is the nucleus, or productsimilarto definition I.03, individual,of a class, to that class. It is structurally relation. Theorem I.56 relationreplacingthe discreteness with the part-whole reveal that forany not-nullclass of individualssuch and subsequent theorems that some one individualis a commonpart of everymemberof the class, there is at least one individualwhichis a nucleus of that class; and that no class has morethan one nucleus,so that we may speak of thenucleusof the class. The concepts of fusion and nucleus just definedare not strictly Boolean concepts,since the Boolean concepts operate withinone logical type, whereas
6 in nor since it is alreadydefined our logistic Identityis not defined taken as primitive vehicle,Principiamathematics. Had it been desirable to developthiscalculusin isolation as of from othertreatments logistic,identitycould have been defined mutual part-whole, I.315: to whichit is equivalentby theorem

x=

y. -

. x < y. y < x

seemedundesirablebecause of the uses, as in Part III below,that This isolation,however, we intendedto makeofthecalculus.

48

HENRY S. LEONARD AND NELSON GOODMAN

relatingconceptsof one type with those of these relationsare heterogeneous, to however, the sums and productsof the next highertype. They correspond, *40.01 and *40.02. But whereas classes defined in Principia mathematics, are sums and products definedin Principia mathematics only applicable to classes of classes, the notionshere definedare preciselyapplicable to classes of by individuals. It is, furthermore, means of these non-Booleanrelations that Boolean operationsof additionand multiplication are analogues to the strictly definedin I.06 and I.07 (below). or the I.05 introduces Universe, the universalelement,U, and I.08 Definition the defines operationofnegation: I.05
U =Df Fu'V

I.06
I.07 1.08

X + Y =Df Fu'(txu ty)


Xy =
Df

Nu'(tx u ty) Fu'"(y

-x

=Df

X)7

of Throughthe introduction the non-Booleanfusionand nucleus,propositions state having no Boolean analogues appear in our calculus. Such propositions the Boolean operations,and more generallaws, of an obvious kind, governing are, in fact,deduced as fromthem the analogues to the Boolean propositions more or less direct corollaries. Yet although these general relationsof individuals to classes have herebeen used to definethe Boolean operationsand the definitions could have been given: universalelement, independent I.05'
U
=Df Y

(X)((y)
=Df (7Z)(W <

y<X)
Z .Z . W < Z = W W X . W TLY)

.06'
I.07'

x +

Xy =Df

(Z)(W

X . W < Y)

I.08'

-x

Df (Z)(Z

l X. X+

U)

would, of course, Replacementof I.05 to I.08 by these alternativedefinitions in alterations the postulates. It would also resultin the elimination necessitate non-Booleanpropositions. from calculusof all the highly the important Of the three postulates,the firstallows us to assert the existenceof some individualwhichis the fusionof a given class wheneverthat class is not null; the second relates the discretenesscalculus to identity,already definedin while the thirdin effect states a generalproperty the of Principia mathematica; primitiverelation: I.1 (3x) .x ea.
D.

(3y) .yFua8

7The present formof this definitionhas been suggested by the corresponding definition in Dr. A. Tarski's appendix to Woodger, op. cit. 8 Le?niewski employs only two postulates. One is identical with our I.13 expanded in terms of the primitive relation; the other asserts both the existence and uniqueness of the fusion of any (non-null) class. Our postulate i.1is weaker, since it asserts only the existence of some such individual but not its uniqueness. Accordingly we require also postulate I.12.

CALCULUS

OF INDIVIDUALS

AND ITS

USES

49

I.12 I-13 I.3

<y

y<x

X=y

x o y _-(x I
X<Y.Y<z x<z

are: Someillustrative theorems I.31


I.325 I.326 I.328 I.327

x <x
'(X << x)
x <<Y D

I.e., "part-whole" a transitive reflexive is and relation. It is non-symmetrical.


(Y <<x)

x << y . y << z . D . x << z x << Y . Y < z . D . x << z

I.e., "proper part"is an irreflexive, asymmetrical, transitive but relation. Theorems like 1.327may be proven every for possible permutation "part" of
and "properpart" excepttheone "x < y.y< z. D.x<<z.

I.331 I.332 I.333 I.53 I.55


I.556

xo y y o x x < y 3Xoy xo x (3x) . x e a . .E! Fu'a


D Fu'a

E! Fu'a . D . a C D Fu'a < Fu'


E! Fu'a . D . a =

= Fu'j

Propositionsassertingconditionallythe converse implicationsof those in the consequentsof I.55.556 are not theorems the system. Many distinctclasses of may have the same fusion. For example,let a be the class of tables, #be the class of table-tops,and 'y be the class of frames(includinglegs, drawersetc.) of tables. It is plain that a n ( u)

classes and no member one is a member the That is, a and ,3u yare distinct of of other. Yet theyisolate the same part of the total universe: Fu'a = Fu'(Q u ey) in for They differ the mannerof subdivisionthat theyprescribe that part. the Sample theorems concerning existenceand uniquenessof the nucleusof a class and the negate ofan individualare: I.56 I.57 (p' < "a) xNua.zNua A.D(3x).xNua D. =D z

50 I.58 I.59 I.6 I.62 I.66

HENRY

S. LEONARD

AND NELSON

GOODMAN

(p' < ") (3y) . y E!x+y x + y=y

# A

D E! Nu'a

x.

El-x

fusionswe have: concerning As special cases of generaltheorems + x

(x + y) + z = x + (y + z)

Theorems analogous to the above may be demonstratedfor "xy", but oniy conditionally upon the assumptionthat "x o y", or,what is equivalent,"E!xy". conditions. For Analogues to DeMorgan's formulaappear under interesting example, I.85 xoy .x $ U . y # U. D .xy=-(-x + -y)

the logic,the calculusofconcepts III. Besides supplementing body ofsymbolic withcertainrelational of lowesttype equips us to exhibitand deal efficaciously or whichare oftenignored misunderstood, sometimes thedetriment to properties like of constructional undertakings Carnap's Logischer Aufbauder Welt.9 the Consider,forexample,the relation"met with"; we cannotdefine ordinary that threeor morepeople all met together to meaningof propositions the effect that everypair met; foreach pair may have metseparately, without by requiring all threeever having met together. Or again, Johnand James may be lodgeand John and Arthurbe brothers,and James and Arthurbe lodge-brothers, withoutall threebeing brothersin any one lodge.'0 Even the lodge-brothers, that a color,C, is, as we customarily say, "at place P commonplaceproposition that C is at place P and at time T. at time T" is not impliedby the proposition in In each of these cases we encountera relationwhichis significant varying as dyadic, triadic,tetradic,and so on. The relation"met with", for degrees; and indeed example,may obtain betweentwo people, among three,amongfour, number. Such a relationwithoutany fixeddegreemay be called a among any "multigrade"relation." Customarymethodsof treatingthe logic of relationswill admit of no simple of introduction multigraderelations,for these methods presume that an exof haustive and exclusiveclassification relationscan be set up in termsof their definite will, providefortheir interpretations however, degree. Two alternative in introduction a mannerconsistentwith this presumption. Either they may the of as be construed each a seriesof relations, successivemembers whichhave as or takingclasses degrees, theymay be construed predicates successively higher for of variousmagnitudes theirarguments. is The secondinterpretation strictly applicableonlyin the case ofa multigrade
I Berlin, 1928. See especially sections 67-93, pp. 108-120. This work is subsequently referredto as the Aufbau. 10 The following discussion of the problem here illustrated is that referred to by W. V. Quine in Relations and reason, Technology review, vol. 41 (1939), pp. 325, 327n. 11A unigrade relation is a relation of any one degree; a multigrade relation is one having degrees. at least two different

CALCULUS OF INDIVIDUALS AND ITS USES

51

relationthat is thoroughly symmetrical. In othercases, it mustbe modified in the mannerindicatedbelow (p. 53). Since, however, relationsthat we shall the first considerare all of themthoroughly symmetrical, adopt this interpretawe tion in the sequel. But whichever view of multigraderelationsis preferred, notable feature the of the examplespreviously given,althoughit does not attach to all multigrade relations any means,is that a givendegreeof the relationor predicatecannot by be reduced,by the use of none but the commonly recognized logical devices, to its lowerdegrees. It is wherethisreduction impossible is that a crucialproblem for confronts constructionalist; considerations economy, the of together withthe special conditionsof his problemin hand, will ordinarily cause the construcof tionalistto rejectas primitives his systemany predicatetakinganything other otherthan individualsas arguments, any wholehierarchy relations, of and any of uppermost member such a hierarchy (shouldit be finite) whenthe identification of that upper limit requires postponementof the formal development of pending the detailed investigation contingent mattersof fact. Thus he is with a dilemma: if multigrade confronted relationsbe admittedwithoutinterpretation, thenthe standardlogic of relations, developedin termsof a classification of relationsaccordingto degree, is inapplicable to them,but if they are admitted by using eitherinterpretation suggestedabove, then oftenthey are and cannot be reduced to acceptable primitives. not themselves The mannerin whichthe calculus of individualsmay be enlistedto simplify the primitives needed, to display the connectionbetweenthe different degrees in of the relationor predicate,and to expressthe distinction meaningbetween and sayingthat each two saying,forexample,that threemen all met together met severally,can best be illustratedby a slightlymore complicatedexample than any of thoseyet given. Suppose we have as elementsa set of threecolumns,each coloredwith three bands, as picturedin the accompanying diagram,and suppose that the relation S is such that "xSy" means that in some one band-lower, middle,or upperthe two entitiesx and y are identicallycolored (in the sense that no color in from that band eitherin x or in y is different any one colorin that band in the other). It is clear that S is a relationlike those already considered;that we may have threecolumns,like the ones pictured,such that aSb, bSc, and aSc, even thoughall threecolumnshave no singlecolorin any one band.

o
M a

L
H b

L
M c

The capital lettersrepresent distinctshades of color

52

HENRY

S. LEONARD

AND NELSON

GOODMAN

of However,nothingin our specification S preventsit fromtakingas relata, not only single columns,or elements,but also those entitiesthat are sums of the elements;the expression"xSy+z" has the perfectly clear and unambiguous meaningthat x and y+z are identicallycolored,in the sense described,at some level. In the illustrationgiven, therefore, "aSb+c" is false, for at whatever level we look, eithera and b+c have entirelydifferent colors,or else a is unicolored while b+c is bi-colored,and the conditionfor the holdingof S is not satisfiedin eithercase. The proposition"xSy+z" will thus hold only if x and y+z-and, therefore, y and z-have a single, identical color at some one x, level; and accordingly the triadicdegreeof the relationmay be definedby the The principle involved is generally extensible to any function "xSy+z." degree;hence we may definethe class-predicate, applicable to classes having S', numberof members, follows: as any finite
S'(C) = Df

# A

$- A .

n 'y = A

. A

u -y C a . Do,, . Fu'j S FuOy

That is, "S'(c)" means that for every two discretesubclasses, if and fy, a, of the fusionof ,3 has the relationS to the fusionof fy. Because the particularrelation,S, chosenhappensto have a certainproperty later to be definedand called "interdissectiveness," expressions"xSy+z," the "ySx+z," and "zSx+y" are all equivalent,and are equivalentto "S'({x,y,zj)." we Taking advantage of this property, can simplify our general definition to read as follows:
S'(C)
=Df

(3,O)p

i n y= A

u fly a =

Fu'i3 S Fu'-y

That is, "S'(a)" means that the relation S holds betweenthe fusionsof some two mutuallyexclusiveand exhaustivesubclasses of a.12 The meaning,then,of saying that x has the relationS to y and z together, not merelyseverally,is given by the expression"xSy+z"; and the same treatment is applicable to the other relationsthat we noted earlier. That a man Smith met with Jonesand Brown togethermeans that he met with the entity whichis the sum of the two. The sum will not be a person,of course,but is a whole. For a colorto be "at a place at a time" definable thoughdiscontinuous is forit to be' at that entitywhichis comprisedof both, the place plus the time, or in otherwords,at the place-time. Likewisefora man to be a lodge-brother of two otherstogether forhim to be a lodge-brother theirsum. is of But caution is necessary here: the suggested treatmentmay be employed only when the relationin question is one, like our S, that takes as relata not but also sums of theseelements, atomic elements, merely and the interpretation of the relationmust be constantirrespective the particularrelata satisfying of
12 The twodefinitions cases in whichthecardinality just proposedare equivalentonlyfor of theargument, is greater thanorequal to two. The first a, definition makes"S'(a)" true of whenthe cardinality a is less than two,whilethe secondmakesit false underthesame circumstances. This difference, restsupon trivialcases that may be decided by however, of in considerations convenience dealing withthe particularsubject-matter and problem in hand; and eitherdefinition may be easily adjusted to accord withwhateverdecisionis made.

CALCULUS

OF INDIVIDUALS

AND

ITS

USES

53

it in a particularcase. If, for instance, "lodge-brother" understoodas a be relationholdingsolely among individualpersons, could not speak of a man we of being the lodge-brother the sum of two otherpersons. We should thenhave to employ as our primitivea broader relation,say "lodge-affiliate," which is not thus restricted in termsof whichthe narrower and relationcould be defined. The general nature of the technique employed in the precedinganalyses mightbe describedas follows. Given foranalysis a multigrade relation,its dyadic cases are abstractedand taken as a primitivedyadic relation. The multigraderelation itself,representedin the illustration givenby S', is thendefinitionally just as introduced the predicateofa class. No restriction put upon the cardinality the argumentis of classes that satisfy it. The definiensof the multigradepredicate involves references the primitive to dyadic case of the relationin question. But many in the definiens this dyadic case are describable of the arguments assigned to only by the use of conceptsdeveloped in the calculus of individuals;generally theyare sums of individualsor fusionsof classes of individuals. (The generaltreatment just describedwould have to be slightly in modified the case of a multigrade relationthat is not whollysymmetrical. "To murder" and "to annoy" are multigrade, inasmuchas one personmay murder annoy or another,or several personsmay cooperate togetherin these actions; but not inasmuchas interchanging whollysymmetrical, termsdesignating murdered the man and one of the accompliceswill not generally resultin a proposition equivalent to the original. All multigraderelationsare, however,at least partially symmetrical. For example,arguments the designating accomplicesin a murder may be interchanged willwithout at the affecting truth-value the propositions of under consideration. Such multigraderelationsmay be definitionally introduced as heterogeneous relationsbetweenthe class of the interchangeable individuals and the others.) That our generaltechniqueenables us to introducemultigrade relationsin a mannerconformable with the guiding principlesof constructionalism already suggested(on page 51), is due to the fact that a+b is a conceptof the same logical type as a and b themselves, and moregenerally that the fusionof a class, a, is a conceptof the same logical type as the members a. Thus "aRb+c," of and "Fu'aRFu'#3"expressdyadic relationsof individuals;they are expressions of the form"xRy" in whichthe individualsconcerned to happen to be referred by moreexplicitdescriptions. Therefore generaldefinition, example of our for the predicateS', has as its definiens logical function a strictly a of dyadic relation of individuals. It may now be noticedthat if the generaltechniquejust outlinedhad been used in defining "quality-class"in the systemof Carnap's Aufbau,the gravest a defects theearlyconstructions of would have been avoided. For theabstraction of these quality classes is supposed to be accomplishedthroughthe use of a relationof similarity that is, in all relevantrespects,like the relationS of the preceding illustration. And the defectsin the construction consistin, or result from,mistakenlysupposing that a class of thingseach memberof which is similarto each otheris a class of thingswhichare all similar.

54

HENRY

S. LEONARD

AND NELSON

GOODMAN

What our procedureaccomplishesis the analysisof any such complicatedand special conceptas S' into a much simplerspecial conceptand easily mishandled which is apart fromany particularsystem the generalrelationof discreteness, relationsarises. The proba and is applicable wherever problemof multigrade and of the colored of lems of the men who met together, the lodge-brothers, columnsare met in each case by employinga special simple relationtogether with the general discretenessrelation. To adopt a complicated predicate of in classes as primitive each case would be to foregoan analysis whichdiscloses the identityof the logical problempresentin all these cases. problemsof the general charThe details of the solutionsof constructional will vary with the specific properties possessed by the acter of those illustrated that the calculus of individuals relationsin question,especiallythose properties and define. The relationsso far consideredhave all assists us to comprehend in that defined D5 below. been non-agglomerative, is, have not had the property of This was the featurethat made impossiblethe definition the higherdegrees of any of these relationsin termsof the lower,so long as the atomic elements alone were considered. On the other hand, all these relationswere also not merely dissective-conformingto D6-but were even interdissective-conto forming D7; and it was forthisreasonthat "aSb+c," "bSa+c," and "cSa+b" were equivalent. I of and may properties relationsand predicatesmay be defined, Many further prove importantin connectionwith other problems. A few of the more inare of teresting these properties definedas follows: D1 D2 R is internal x R y if
DX

vx o y
Y

R is external x R y Dxv x I if

internaland externalrelaA good deal of the traditionaldiscussionconcerning of tions mightbe clarified consideration these two simple,but systematic, by definitions. The relationof a part to a whole and of a quality to the object that it qualifiesare examples of internalrelations. Things with no common and withdiverslocationsin space and time are in generalexternally properties related. D3 D4 R is expansive x R y x,,,.x R y + z if R is cumulative x R y . x R z . bXy, a.x R y + z if and overlappingare expansive relations. Part-whole but the converseis not always true. DisAny expansiverelationis cumulative, are cumulativewithout for creteness, example,and the converseof part-whole beingexpansive. D5 ifxRy.xRz.yRz. y R Ris agglomerative z,, xRy+z Every cumulative-and hence also everyexpansive-relation is agglomerative, and of meetingthat but the converseis not true. The relationsof similarity and we analysedin the earlierpages of this articleare non-agglomerative, hence

CALCULUS

OF INDIVIDUALS

AND

ITS

USES

.55

and non-cumulative non-expansive. The converseof "properpart of"-defined above in I.011 (p. 47)-is agglomerativebut not cumulative and hence not expansive. D6 if R is dissective x R y . z <y. y .x R z The converse of the part-wholerelation is dissective. Also the relation S, elementson that theoriginal analysedabove, is dissective the tacit assumption a, b, and c in the illustration-are to be construedas unanalysableinto proper parts. D7 if R is interdissectivex R y . w < x + y . z < x + y .3,..w Rx relationis dissective,but the converseis not always true. Every interdissective The relationS, analysed above, is interdissective. The converseof part-whole is dissectivebut not interdissective. all by It wouldbe possibleto extendthislistofdefinitions defining thevariants that could be distinguished throughthe possible nonof the above properties if symmetryof the relation involved. For example, R is counterexpansive xRyD.,1.5x+zRy. There are seventeensuch variants of the propertydefined the in D5. When the relationis symmetrical, variants will be equivalent to are definedabove. Also certainof these properties only special the properties applications to relations of propertiesthat might be definedfor predicates generally;forexample, 0 is expansiveif 0(y): Dy.,(y+z). IV. The utilityof the proposedcalculus of individualsis by no means limited in to its usefulness treatingthe problemsconsideredin this paper. Because it because its primitiveand definedideas closely parallel is simple in structure, relation (namely, "partbecause its major non-symmetrical intuitivenotions, otherthan membership relationships because it introduces is transitive, whole") "nucleus of") between individuals and classes, and (namely, "fusion of" and manyvaried entitiesby means especiallybecause it providesmeans fortreating and of individualsis a powerful of conceptsof a singlelogical type,the calculus the In addition it performs for constructionalwork. expedient instrument metaimportantserviceof divorcingthe logical concept of an individualfrom and interphysicaland practicalprejudices,thus revealingthat the distinction and relationof classes and wholesis capable of a purelyformaldefinition, that both concepts,and indeed all the concepts of logic, are available as neutral analysisof the world. Then, forexample,it becomes toolsforthe constructional clear that the practice of supposing that thingsare what the x's and y's of denominate and that qualities are necessarily to be Principia mathematics as interpreted logical predicates thereof,rather than vice versa, is purely a and realistas to what actual matterof habit. The disputebetweennominalist entitiesare individuals and what are classes is recognizedas devolvingupon necessity. convenienceratherthan upon metaphysical mattersof interpretative
DURHAM, NORTH CAROLINA AUBURNDALE, MASSACHUSETTS

Вам также может понравиться