Вы находитесь на странице: 1из 15

Journal of Philosophy, Inc.

The Individuation of Action Author(s): Alvin I. Goldman Source: The Journal of Philosophy, Vol. 68, No. 21, Sixty-Eighth Annual Meeting of the American Philosophical Association Eastern Division (Nov. 4, 1971), pp. 761-774 Published by: Journal of Philosophy, Inc. Stable URL: http://www.jstor.org/stable/2024949 . Accessed: 27/12/2013 01:46
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.

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

http://www.jstor.org

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

76I

analysis of the conquestion requires,among other things,a fuller than I knowof. However,it is not clear that cept of a determinable content. A such principlesthemselvesare devoid of informative relevant featureof the partiallyquoted statementsby Russell and likedeterminables oversomething Kemenyis thetacitquantification whichis deceptivelyconcealed by such littlephrasesas 'one definite way' and 'whatit will be'. care to fitthem When these notionsare analyzed with sufficient forincorporation into a reasonablypreciseexplicationof the thesis it is highlylikelythat we shall discover of Laplacian determinism, a contentwhich,because of its rich and complexontologicalimport and its apparent roots in intuitionsthat many regard as "purely logical," deservesto be called metaphysical.I concludethat at least a substantialpart of the ontologicaland metaphysical burdenof the is borne by its componentclaim thesis of Laplacian determinism that the universeis omnitemporally determinate with respectto at least some determinables. Wesleyan University
KENT BENDALL

THE INDIVIDUATION

OF ACTION

currentphilosophicaldiscussionby G. E. M. Anscombe,1 who asked the question:

into was introduced HE problem of act individuation

moves his arm, Are we to say that the man who (intentionally) the water-supply, poisonsthe inoperatesthe pump,replenishes is performing four actions?Or onlyone? (45) habitants, This is an example of a larger problem,the problem of when, in no solugeneral,act A is the same as act A'. Miss Anscombeoffers tion to this general problem,no general criterionof act identity. But her remarks concerningthis example can be examined with She answersher question as follows: profit. is thisone, actionofhisthatis in question In short, theonlydistinct hisarmup and downwithhisfinger thepump round A. For moving thepump;and, in these operating handleis, in thesecircumstances, and,in these thehousewater-supply; circumstances, it is replenishing it is poisoning thehousehold (46). circumstances,
* To be presented ofthesametitle, December 27, 1971.CominAPA symposium Thomson and Irving Thalberg;see thisJOURNAL, Jarvis mentators willbe Judith The University ofMichigan has respectively. thisissue,pp. 774-781and 781-787, to publish thisarticle length. at itspresent makeit possible provided funds which 1 Intention Basil Blackwell, 1957). (Oxford:

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

762

THE JOURNAL OF PHILOSOPHY

The 'is' hereis clearlythe 'is' of identity;what Anscombeintendsto assert is that the man's movinghis arm up and down = the man's of the watersupply operatingof the pump = the man's replenishing the man's poisoningof the inhabitants. The order in which these putative identities are presented is is aware of this,forshe speaks of the Anscombeherself significant. as forming. fourdescriptions
-

as is introduced ... a series,A-B-C-D, in whicheach description ofthefollowing independent one,though on theprevious dependent one.2 hereis confirmed by considering The existenceof a naturalordering of the elements. The followingarrangementis a rearrangement clearly jumbled: "Replenishingthe water supply is (in these circumstances)operatingthe pump; and it is (in these circumstances) moving poisoningthe inhabitants;and it is (in thesecircumstances) an asyminteresting in particular, Note, arm and down." up his up his arm moving to say that metry.Althoughit is fairlynatural to the it less natural water-supply, is and down "is" replenishing "is" (in thesecircumstances) the water-supply say that replenishing movinghis arm up and down. fact for The existenceof a natural orderingis an embarrassing if we grant explained The is readily ordering Anscombe's position. in a certain which stand the distinct acts designate that descriptions relation to one another. But if they all designate the asymmetric thereare natural Admittedly, same act, whencearises the ordering? the that name same object. A natural among descriptions orderings '1 ordering exists among the descriptions + 1', '2 + O', '3 - 1', '4 - 2', etc., thoughtheyall designatethe number2. In Anscombe's is imposednot case, the ordering unlikethe numerical case, however, the If the man's world. rather but by themselves, by thedescriptions be a different would required. ordering behaviorhad been different, Had his arm been restingon the pump, he mighthave moved his arm up and down by operating the pump (with his other arm). sort of orderingamong the And to express this fact, a different would be appropriate.There is reason to suspect,theredescriptions of the are not in factdescriptions that Anscombe'sdescriptions fore, verysame action,that the 'is' it seems naturalto employhereis not the 'is' of identity. The use of the preposition'by' provides additional, and more
2 Ibid.,p. 45. Anscombe withtheman'sintentions. seemsto associatethisseries even if the man did not have all the But the same serieswouldbe appropriate to him. ascribes Anscombe intentions

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

763

conclusive,evidence for the same point. It is true to say that the the water supply,that man poisons the inhabitantsby replenishing he replenishesthe water supply by operatingthe pump, and that he operatesthe pump bymovinghis arm up and down. As used here, the preposition'by' seems to expressa relationthat holds between the water supply and an acts, e.g., between an act of replenishing act of operatingthe pump (in that order). Now ifAnscombeis right in claiming that the man's operatingof the pump is identicalwith his replenishing of the water supply, then any relation that holds betweenthese acts in one directionmust hold betweenthemin the opposite direction.Moreover,the relationmust hold betweeneach of these acts and itself.But thoughit is true to say that the man replenishesthe water supply by operatingthe pump, it is false to the water supply. say that he operates the pump by replenishing (Similarly,it is false to say that he moves his arm up and down by operating the pump, or that he replenishesthe water supply by poisoningthe inhabitants.) Moreover,it would be odd to say that the man operatesthe pump byoperatingthe pump. The by-relation, and irreflexive relation.Since no then,seems to be an asymmetric such relation can hold between a given thingand itself,we must conclude that the acts in this example are not identical. with There may be an initial tendencyto confusethe by-relation in for cases the between the causal relation, holds many by-relation a pair of acts in virtueof a causal relationholdingbetweenone of theacts and one ofits consequences.This confusion mustbe avoided. Suppose that I push the buttonand thiscauses thebell to ring.There is a causal relation that holds between my pushingof the button holds betweenthisact and but no by-relation and the bell's ringing, this consequence. In virtueof the causal relation,though,it is true to say that I ring the bell by pressingthe button. Thus, the byrelation holds between my ringingof the bell (i.e., my causing the bell to ring) and my pushingof the button.The causal relation, however,does not hold between these two acts of mine (in either direction). depend upon causal Althoughmany instancesof the by-relation holds betweenJohn's relationships, many do not. The by-relation signalingfora turnand John'sextendinghis arm out the window, of his promise and John's and it holds between John's fulfilling of thesecases, though, of the book. In neither does the byreturning relationobtain in virtueof a causal consequence.The existenceof a certainconventionmakes it true that Johnsignalsbyextendinghis arm, and the existence of a certain background condition (viz.,

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

764

THE JOURNAL OF PHILOSOPHY

that Johnpromisedto returnthe book) makes it truethat he fulfills his promisebyreturning the book. It is instructiveto think of the by-relationin diagrammatic terms.In Anscombe'scase we can capture the forceof the by-relation by drawinga columnof circles,the lowest of whichrepresents the man's moving his arm up and down and the upper nodes of whichrepresentthe other acts. (See Figure 1.) A vertical line (or, later, a diagonal line) indicates that the upper circlerepresents an act that bears the by-relation to the act represented by the lower circle. As the diagram suggests,the by-relation is not only asymmetricand irreflexive, but also transitive.
( theman'spoisoning oftheinhabitants theman'sreplenishing ofthewater supply theman'soperating ofthepump theman'smoving hisarmup anddown

Figure 1 Although Anscombe's example can be representedwith a single of the sortsof relationsthat columnof circles,a completetreatment hold betweenacts requiresmorecomplexdiagramming. Specifically, we must admit the possibility of tree-shaped diagrams, with branches going off in different directions. Suppose that Boris the squeezes his finger, therebypulling the trigger, therebyfiring Pierrefrom gun,thereby killing Pierre,thereby preventing divulging the party's secrets, therebysaving the party from disaster. By killingPierre,he also drives Pierre's lover to suicide. The by-relaheredo not permita singlecolumnof acts. AlthoughBoris tionships drivesthe lover to suicide by killingPierreand saves the partyfrom disaster by killingPierre,he does not save the party fromdisaster by drivingthe lover to suicide; nor does he drivethe lover to suicide by saving the party fromdisaster. This case is shown in figure2.
Boris'ssavingthe partyfrom 0 disaster Pierre Boris'spreventing from theparty's secrets divulging Boris's ofPierre killing Boris's ofthegun firing Boris's ofthetrigger pulling hisfinger Boris's squeezing

Boris's Pierre's lover to driving suicide

!
Figure 2

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

765

These simple examples place in clear reliefthe most important kinds of relationships that are ignoredby Anscombe's view on act identity.Not only does her treatment of such acts neglect the of the by-relation, asymmetry but it also ignoresthe distinctiveness of the relationthat obtains betweena pair of acts that bear the byrelationto a commonact, thoughnot to each other. The problemconfronting Anscombe's view on act individuation is equally a problemforotherwriters. Donald Davidson's treatment of particularcases, for example, is substantiallythe same as Anscombe's. In "Actions, Reasons and Causes" Davidson writes: I flipthe switch, turnon the light,and illuminate the room.Unbeknownst to me I also alerta prowler to thefactthat I am home. HereI do notdo four things, butonlyone,ofwhich four descriptions have been given.3 Unlike Anscombe, however,Davidson has buttressedhis position with considerable argumentation. Moreover, in another paper, "The Individuation of Events,"4 Davidson proposes a general of act individuation.(Strictlyspeaking,it is a criterion criterion of event individuation, but it is intendedto hold foracts as well,since actions are a species of events.) The generalcriterion is this: (1) Events are identical if and only if they have exactly the same causes and effects. I do not believe that sameness of causes and effects is a sufficient conditionforact identity.But I have no quarrelwiththe claim that it is a necessarycondition;indeed,assumingthat beinga cause of E ofE are genuineproperties, and beingan effect its beinga necessary condition follows from the indiscernibility of identicals. What I wish to stress,however,is that Davidson's treatment of particular on his own generalcriterion. cases of putative act identityfounders Apparently,many pairs of acts he would class as identical do not have exactlythe same causes or do not have exactlythe same effects. Let us look firstat effects.Consider Boris's act of pulling the and his act of killingPierre. Then consider the event that trigger consists of the gun's firing. Clearly, Boris's pulling of the trigger causes this event. But does Boris's killingof Pierre cause it? Certainly (2) Boris's killingof Pierrecaused the gun to fire seems to be false; at any rate, it sounds extremelyodd. If it is,
' This JOURNAL,LX, 23 (Nov. 7, 1963): 685-700,p. 686. 4In N. Rescheret al., eds., Essays in Honorof Carl G. Hempel(Dordrecht: Reidel,1969),pp. 216-234.

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

766

THE JOURNAL OF PHILOSOPHY

indeed, false, then by appeal to (1) we must conclude that Boris's pullingof the trigger 5 Boris's killingof Pierre. has already been hinted at: A possible reply to this difficulty although (2) sounds odd, it mightbe argued,it is nonethelesstrue. But how can thisreplybe justified?A clue is providedin a passage is to describe whereDavidson says: "To describean eventas a killing it as an event (herean action) that caused a death."5 This suggests the view that (3) Boris's killingof Pierre is to be paraphrasedas (4) the action of Boris's that caused Pierre's death then presumably(3) and (4) would If this paraphrasewere correct, It could next be argued that the referent of (4), be co-referential. Since Boris's and hence of (3) as well,is Boris's squeezinghis finger. caused the gun to fire,it would then be consqueezing his finger cluded that (2) is true. Can it be established that (4) is a correct paraphrase of (3)? It mightbe noted, with this in mind,that sentences (5) Boris killed Pierre and (6) There is an action of Boris's that caused Pierre's death are "virtually" equivalent. (I say virtually equivalent because one can cause Pierre's death withoutkillinghim,e.g., by gettingsomeone else to shoot him. But for our purposes this point can be ignored.) Now if (3) and (4) were derived from(5) and (6) by the therewould be reason to thinkthat they are same transformation, paraphrases of each other, or at least co-designative.However, they are not so derived: although (3) is a gerundialnominalization of (6). Moreover,even of (5), (4) is not a gerundialnominalization is unsound.Not all gerundial nominalizations the suggestedprinciple of equivalent sentencesare co-designative.For example, although 'EffectE occurred'and 'Somethingcaused E to occur' are equivalent, their respective nominalizations: 'Effect E's occurring'and There is 'Something's causing E to occur', are not co-designative. not much reason to believe, then, that (4) is a good paraphrase of Of course,if (3) cannot (3), nor that (3) and (4) are co-referential. be construedas (4), one wants to know how it should be construed. I shall returnto this question below. that identical actions have Let me turnnow to the requirement the same causes. Davidson clearly wants to regard the following
5 Ibid.,p. 229.Alsosee Davidson'spaper"Agency," inR. Binkley, R. Bronaugh, andReason(Toronto:University Press,1971), Action, and A. Marras, eds.,Agent, pp. 3-25.

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

767

simultaneous actions as identical: John's singing,John's singing loudly,and John'ssingingoffkey. This seemingly innocuousview, however,runsinto difficulties. John'ssingingloudly,let us suppose, is partlycaused by his beingangry,but his beingangrydoes not at all cause his singingoffkey. Similarly,his singingoffkey is partly caused by his having a sore throat,but his having a sore throatis not at all a cause of his singingloudly. Accordingto (1), therefore, John'ssingingoffkey # John'ssingingloudly. Two repliesare open here. First,one mightsay that, contraryto appearances, John's singingloudly is caused by his having a sore throat,and his singing off keyis caused by his beingangry.How can thesecontentions be supported, however, withoutrelying on the disputed identity of the acts in question (or on the dubious assumption that John'sbeingangry = his havinga sore throat)? Secondly,one mightreplythat the effect of John'shaving a sore throatis not the "entire" action of singingoffkey, but simplyone aspect, featureor property ofit, viz., its beingoff key.Similarly, the effect of his being angryis not the entireaction, but simplyits being loud. This reply raises importantquestions concerning the conceptionof an action. What thingsare "constitutive" of an action and what thingsare mere aspects or properties of it? If somethingcauses the direction of an arm movement, does it cause the action itself,or merelyone featureof it? However one answers this question, one is forcedto admit that some of the terms in causal relations are things that consistin something's havinga certain property: eitherJohn'shaving the propertyof singingloudly,or his action's having the property of being loud. Once the need forsuch particularsis recognized, the conception of an act I wish to recommend receives additional support.But this runs ahead of the story. Let us mention one final problem for the Anscombe-Davidson mode of act individuation.This is the temporal problem,whichhas been raised by Lawrence Davis' and by JudithJarvis Thomson.7 Suppose that Johnshoots George at noon but George does not die of his wounds until midnight.It is true that John killed George and that he killed him by shootinghim; but is the killingthe same as the shooting?Though it is clear that George'sdeath occurstwelve hours after John shoots him, it seems false to say that George's death occurs twelve hours afterJohn kills him. But if the death followsthe shooting, but not the killing, by twelvehours,the shooting and the killingmust be distinct.
6

"tIndividuation ofActions," thisJOURNAL, LXVII, 15 (Aug.15,1970): 520-530. "The Time ofa Killing," ibid.,LXVIII, 5 (March 11, 1971): 115-132.

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

768

THE JOURNAL OF PHILOSOPHY

sort of problemin "The Individuationof Davidson discusses thlis Events." He pointsout that we may easily knowthat an action is a pouringof poison withoutknowingthat it is a killing; so we may be unpreparedto describe the action as a killinguntil the death one. Even afterthedeath occurs.But theproblemis not an epistemic has occurred,and we knowit has occurred,it still sounds wrongto say that it occurredtwelve hours afterthe killing.Davidson goes on to suggest that there may be a tendencyto confusean event withan event describedin termsof a describedin termsof an effect of "terminalstate" (e.g., 'his painting the barn red', the referent paintingthe barn red). But which is not over until he has finished when what groundsdoes one have forsupposingthat we are confused we thinkthat John'skillingof Georgeis not over untilGeorgedies.? There is a conflicthere between a pre-analyticintuition and a theory.Though it is possible to protect the theoryby calling the intuitionconfused,it is just such intuitionsthat must be used in testingthe theory. Moreover, the intuitionthat John's killingof Georgedoes not precede George's death by twelvehoursis just the sort of intuitionthat sheds doubt on the view that phrases like (3) can be paraphrasedaccordingto the model of (4). If 'John's killing of George' just means "the act of John's that caused George's death," why does it strikeus as false to say that John's killingof George preceded the death by twelve hours? Why does this seem false even when we know that the shootingcaused the death and that it preceded the death by twelve hours? reasons have now been given for regardingthe AnsSufficient combe-Davidson pattern of act individuationas inadequate. But although theirparticularidentityclaims are mistaken,or at least highly questionable, it does not follow that Davidson's general criterionfor act identityis incorrect.I have indicated my doubts conditionfor act identity; but even that (1) provides a sufficient be whollysatisfactory. Ideally, a not it if (1) were correct, would that be "correct"-if merely not should criterionof act identity a criterion as law Leibniz's with content were enough,why not be is that it A further desideratumof such a criterion of act-identity? of an act. Although lay bare the nature, or ontological status, Davidson's causal criteriontells us somethingof ontological importance about actions-viz., that they are terms in causal relations-there is more to be said. of a fullyadequate criterionof identity,then, The formulation of the ontologicalstatus of an act. requiresa betterunderstanding I suggest we begin by considering To attain this understanding,

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

769

ordinary action sentences. A satisfactoryanalysis of action senwhat tences, I believe, requiresthe notionof an act type (roughly, of thisnotion othershave called a "generic" action). The usefulness derives fromvarious sources. Frequentlywe wish to say that the same act is performed on more than one occasion, e.g., that John the same act did the same thingas Oscar, or that Johnperformed today as he did yesterday.The act in question here cannot be an "individual" or "concrete" act, foran individualact cannot occur at different times,nor can two personsbe agents of the same individual act (neglectingthe possibilityof collective actions). The notionof an act type is also needed in talkingabout ability.If I say that there is an act that John, but not Oscar, has the ability to I do not assert the existenceof any "concrete" action; for perform, an act that Oscar is unable I do not implythat Johnhas performed theexistence is best construed as asserting to perform. My statement of an act typewhichJohn,but not Oscar, has the abilityto perform. something An act type,as I construeit, is simplyan act property, thatan agentexemplifies. Whenwe say, "Johnweighed170 pounds" or "John was bald," we ascribe to John the propertyof weighing of beingbald. Similarly,I suggest,when 170 pounds or the property we say "John signaled for a turn" or "John killed George," we of signaling ascribeact properties or act typesto John:the property fora turnor the property of killingGeorge.To ascribe an act type it. If Johnand Oscar perto someone is to say that he exemplified form thesame act, theyexemplify the same act type.If Johnhas the the property abilityto wigglehis ears, he has theabilityto exemplify between of wigglingone's ears. Admittedly,there is a difference an act. This a property, in general,and performing exemplifying I believe, is to be analyzed in termsof what causes the difference, of the property.If I sneeze as a result of the usual exemplifying the property of sneezing,but I do not perform causes, I exemplify an act. If sneezingis undermy voluntarycontrol,however,and if I exercisethis controlby sneezingon purpose,then I have performed an act of sneezing. How adequate is the approach to action sentences that I am recommending? Davidson has argued that an analysis of the logical formof action sentencesmust account forthe fact that 'Sebastian strolledthrough the streetsof Bologna' entails 'Sebastian strolled'.8 each of thesesentencesascribes Accordingto my analysis,however,
8This pointwas first raised by Davidson in "The Logical Form of Action Sentences," in N. Rescher, ed., The Logic ofDecisionand Action(Pittsburgh: University Press,1967).

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

770

THE JOURNAL OF PHILOSOPHY

an act propertyto a person. This seems to imply that these senas 'Bs' and 'Ss', where 's' tences are to be parsed, respectively, stands for 'Sebastian', 'Bx' for 'x strolls through the streets of predBologna' and 'Sx' for 'x strolls'. If these are unstructured ensured? icates, however,how is the entailment Recent wvork by Romane Clark' and Terence Parsons'0 suggests how to handle the problem. (A related suggestion is made by Roderick Chisholm.") On the Clark-Parsonsapproach to predicate a predicatemodifier is construedas an operatorwhich modification, maps a propertyexpressedby one predicate onto a new property. 'Slowly', forexample, is an operator which,when attached to the predicate 'x drives', yields a new predicate,'x drives slowly', that expresses the propertyof drivingslowly. Clark suggests that for standard modifiers (excluding "negators" such as 'nearly'), a detachment holds. Let an initial segmentof a principleof predicate a predicate be either a "core" predicate or the result of prefixing modifieroperator to an initial segment. Then any predicate P implies any initial segmentof itselfQ-that is, any statementsP* in the term filling and Q* that result fromP and Q by uniformly positionsare such that P* entails Q*. With this rule of detachment, it is easy to see that the predicate 'x strolled throughthe streets of Bologna' implies 'x strolled', hence that 'Sebastian strolled throughthe streetsof Bologna' entails 'Sebastian strolled'. the logical formof senGiven this technique for representing we may continue to regard such tences with predicate modifiers, sentencesas ascribingact typesto persons.Of course,since strolling propertyfrom the throughthe streets of Bologna is a different propertyof strolling(though perhaps in some sense "constructed from" the latter property),the two sentences in question ascribe the act typesto Sebastian. However,nobody can exemplify different the second at the first of these act types withoutalso exemplifying same time. the notionof an act type,but our main interWe have introduced est is in "individual" acts, i.e., acts that have a particularagent, that occur at a particulartime (or duringa stretchof time), and that serve as termsin causal relations.Obviously these cannot be act types, which are universals. Let us call these individual, or particular, acts "act tokens." Since an act token is standardly designated by a nominalizedformof an action sentenceand since
9 "Concerning 1970): Modifiers," theLogicofPredicate Nois, iv, 4 (November 311-335. 10"The Logic of Grammatical xxi (1970): 320-334. Synthese, Modifiers," 1 "StatesofAffairs Again,"N,ois,v, 2 (May 1971): 179-189.

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

771

an action sentence associated with such a nominalizationasserts it is naturalto view a certainact property, that a personexemplifies of an the designatumof such a nominalizationas an exemplifying act propertyby a person. Thus, John's jumping ten feet is an exemplifying by John of the propertyof jumping ten feet. And by Boris of the property Boris's killingof Pierreis an exemplifying (act type) of killingPierre. Moreover,since the act type of killing it seems Pierre is distinctfromthe act type of pulling the trigger, of the act type of killing natural to say that Boris's exemplifying of the act typeof pulling Boris's exemplifying Pierreis distinctfrom the trigger. the same act Since, in general, the same person can exemplify type on different occasions and since we want to count these exemas distinct,we should incorporatethe temporal element plifyings of an act token. We may say, then, that into our characterization an act tokenis an exemplifying of an act typeby a personat a time (or duringa stretchof time).12 Actually,even this characterization is not quite complete,because a personmay be the agent of two or of the same act type at the same time. If, at more exemplifyings timet,Johnpoints withhis righthand and pointswithhis left,then he is the agent of two simultaneousact tokens of pointing.This the way in which an act problemcan be dealt with by specifying in this case, eitherwith the righthand or with token is performed, will be ignoredin the left.For simplicity, however,this refinement what follows. forthe individuation We are now in a positionto state a criterion theirontologicalstatus: of act tokensthat reflects (7) For any act token A and any act token A', where A is the of 4t' of X by X at t and A' is the exemplifying exemplifying and t = t'. by Y at t', A = A' if and only if X = Y, q = A,,
la My conception of an act token bears a close resemblanceto the account of an (individual) event given by Jaegwon Kim in "On the Psycho-physical Identity iII, 3 (July 1966): 22 7-235, in "Causes Theory," AmericanPhilosophical Quarterly, and Events: Mackie on Causation," this JOURNAL, LXVIII, 14 (July 22, 1971): pp. 426-441, and in "Causation, Nomic Subsumption,and the Concept of Event," forthcoming;and R. M. Martin, in "On Events and Event-descriptions," in J. Margolis, ed., Fact and Existence (Oxford: Basil Blackwell, 1969). The distinction between an act type and an act token is also considered by Charles Landesman, in "Actions as Universals: An Inquiry into the Metaphysics of Action," vi, 3 (July 1969): 247-252. Although LandesAmerican Philosophical Quarterly, fact" man rejects the term 'act-token', his concept of an "action-exemplifying is very close to my notion of an act token. My theory might be brought even closer to that of Kim and Martin by construingact types not as relational properties (e.g., killingGeorge) with objects or patients "built in," but rather as two-term (or n-term) relations (e.g., killing) "with whichan agent and patient jointlyexemplify(or which an agent exemplifies respect to" a patient).

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

772

THE JOURNAL OF PHILOSOPHY

This criterion obviouslyhas the consequencethat many of the acts equated by Anscombeand Davidson are classed as distinct.In light of the problems that were raised for the Anscombe-Davidson approach, this is a welcomeconsequence. yields thatalthoughour criterion It mustbe emphasized,however, a multiplicityof acts where the Anscombe-Davidson approach of the same descriptions yields just one, therecan still be different to by a act token. Like any entity,an act token can be referred expressions.Just as an act token may variety of nonsynonymous John's many properties. it may exemplify have many descriptions, of beingcaused theproperty of thepump (at t) exemplifies operating of causing the inhabitantsto be by a certaindesireand the property of operatthe property poisoned. This act tokendoes not exemplify betweenthe property ing the pump, however.We must distinguish expressedby expressedby 'x operates the pump' and the property is exemplified by John, 'x is an operatingof the pump'. The former not by John'soperatingof the pump (at t). The latteris exemplified the latter is by John's act, but not by John. In my terminology, thepump,' and this of the type,operating equivalent to 'x is a token predicatecan be true only of an act tokennot of a person. a propertyand being a The distinctionbetween exemplifying To say of an act token that is very important. token of a property it is a killing of Georgeis not,on myview,to assertthatit exemplifies the property of causing George'sdeath; rather,it is to say that it is a token of the act type, killing George. Now, although John's the propertyof causing George's shooting of George exemplifies death, it is not a tokenof the act type, causing George's death. There is just one act type of which it is a token, viz., shooting Thus, althoughJohn'sshootingof George is an action of George.13 John's that causes George's death, it is not a killingof George. acts has the virtueofdistinguishing Our criterion of individuation as wellas thecausal in a mannerthat accommodatesthe by-relation, of acts that it and temporal propertiesof acts. The multiplicity countenances, however, may seem ontologically objectionable. many act types at one indefinitely Since an agent may exemplify many simultaneousact time, he may be the agent of indefinitely of tokens. Is this an unacceptable consequence? This multiplicity acts may appear to credita personwithtoo many accomplishments. of an agentis not to be measured But the degreeof accomplishment
13 We mightrelax our requirementto allow John's singingoffkey to be a token key. This relaxation would not of the type singingas well as of the type singingoff imply, however, that John's killing of George is a token of the type, slooting

George.

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

THE INDIVIDUATION

OF ACTION

773

by the number of his acts. Whether someone "gets a lot done" in a day is determinedby whathe does, not by the numberof his sort of objection to the criterionis that it act tokens. A different actions: it implies fails to provide an explicitmethod forcounting many act tokens,no particularnumber the existenceof indefinitely A criterion of individuaof them.This concernbetraysa confusion. tion need not providea methodof counting.One can providea criterion of identityfor patches of red or for pieces of wood (e.g., "occupies exactly the same place at the same time") withoutprovidinga principleforcountingpatches of red or pieces of wood. We do not need a criterion of numerosity of heed Quine's precept,"No entitywithoutidentity."''4 of our analysis may look Althoughthe ontologicalcommitments unattractiveat first glance,its ontologicalvirtuesmustnot be overlooked. Instead of treatingactions (or events) as a primitiveor irreduciblecategory, our account reduces act tokens to persons, act properties,and times. This supports the Aristotelian-Strawsonian ontologyin which substances are primary,and events and states of affairsare derivative. If ontological parsimonyis, as I believe,primarily a matterof the numberof kindsof entitiescountenanced,ratherthan the numberof instancesof a given kind that are countenanced,then the view of action I propose faresverywell on the dimensionof parsimony.Ultimately,of course, an analysis of action cannot be assessed purely in terms of ontological parin theory simony,but must be judged by its general fruitfulness construction. On this count too, I believe, our approach proves its mettle.'5 The mere multiplicity of acts countenancedby our theoryis not so troublesome.It would be a serious flaw, however,if we were unable to account forthe importantunity among the acts in, say, the pumpingcase, a unity which is rightlystressedby Anscombe capable of capturing and Davidson. But our own theoryis perfectly the unitythat one senses here. Instead of conceivingof it in terms of the relation of identity (which has the disadvantage of being we capture it in termsof the by-relaand reflexive), symmetrical tion (or a relation that closely approximates the by-relation). conception, we thinkof thisunityin terms Using our diagrammatic of a single act tree,where each of the nodes on the tree represents eithera "basic" act or an act that bears the by-relation to a single "basic" act.
This pointis made by Davidsonin "The Individuation of Events," op. cit. See myA Theory ofHumanAction(Englewood Cliffs, N. J.: Prentice-Hall 1970).
14 15

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

774

TIE

JOURNAL OF PHILOSOPHY

John'sdeterring a pedestrianfrom the street crossing

0 John'sconvincing his examiner that he is a competent driver

John's hishandwet getting

O
(:) 0

John's a turn for signaling


John's extendinghis arm out the window John'sextending his arm

Figure 3 Strictlyspeaking,the set of distinctacts we want on a singleact tree cannot be ordered by the ordinary"by" locution. According to our criterion of individuation, John's singing (at t) and John's singing loudly (at t) willbe distinctacts; but we would not ordinarily say either that John sings "by" singingloudly or that Johnsings loudly"by" singing.This problemcan be handled by introducing a slightlybroaderrelation,which I have called levelgeneration, under which the ordinary by-relation is subsumed. I have triedto analyze the notionof level generation elsewhere, and cannot reviewit here.'6 I believe,however, that the inclusionof the additionalcases does not upsetourdiagrammatic conception. For example,thepairof acts consisting in John's extendinghis arm and John's extendinghis arm out the windowcan be neatly and naturallyfittedonto an act tree as shown in Figure 3. The analysis I have sketched, then, satisfiestwo fundamental desiderata. First, it slices the units of action thinly enough to accommodate the by-relation. This need has been feltby a number of philosopherswho have drawn a distinctionbetween basic and nonbasic actions, but my theoryallows for a strongerorderingof acts than a meredichotomy. Secondly,we make use of thisordering to introducethe notion of an act tree, and we use this notion to explicate the unity among the diverse acts in, say, Anscombe's case. pumpingcase or Davidson's switch-flipping The University of Michigan
ALVIN I. GOLDMAN

INDIVIDUATING

ACTIONS

the a man replenishing ISS ANSCOMBE had imagined


water supply of a house by operating a pump. She it would be true to say thinksthat,in the circumstances,

16 Cf. chapter II, ibid. of Action, on The Individuation in an APA symposium To be presented of the same title, article on AlvinI. Goldman's 27, 1971,commenting December thisJOURNAL, thisissue,pp. 761-774.
*

This content downloaded from 132.174.255.116 on Fri, 27 Dec 2013 01:46:05 AM All use subject to JSTOR Terms and Conditions

Вам также может понравиться