Cummins- Functional Analysis

8/3/2019 Cummins- Functional Analysis

1/26

Journal of Philosophy, Inc.

Functional AnalysisAuthor(s): Robert CumminsReviewed work(s):Source: The Journal of Philosophy, Vol. 72, No. 20 (Nov. 20, 1975), pp. 741-765Published by: Journal of Philosophy, Inc.Stable URL: http://www.jstor.org/stable/2024640 .

Accessed: 07/01/2012 01:22

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 ofcontent 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
http://www.jstor.org/action/showPublisher?publisherCode=jphilhttp://www.jstor.org/stable/2024640?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2024640?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=jphil


2/26


3/26

742 THE JOURNAL OF PHILOSOPHYtions are just that: assumptions.They have never been systemati-callydefended;generally hey re not defended t all. I think hereare reasons o suspect hatadherence o (A) and (B) has crippledthemost erious ttempts o analyzefunctional tatementsnd explana-tion,as I will argue in sections and II below. In section n, I willbriefly evelop an alternative pproach to the problem.This al-ternative s recommended argelyby the fact that it emergesastheobviousapproachoncewe take care to understandwhy ccountsinvolving A) and (B) go wrong.

III beginthissectionwith a critique ofHempel and Nagel. The ob-jectionsare familiarfor the mostpart,but it will be well to havethemfreshn our minds,fortheyform he backdropagainstwhichI stagemy ttack n (A) and (B).Hempel's treatment f functional nalysisand explanation is aclassic example of the fusionof (A) and (B). He begins by con-sidering hefollowing ingularfunction-ascribingtatement:(1) The heartbeat n vertebrates as the function fcirculating hebloodthrough heorganism.He rejectsthesuggestion hat function' an simplybe replacedby'effect' n the groundsthat, lthoughtheheartbeathas theeffect fproducingheartsounds,his s not its function. resumingB) fromthe start,Hempel takesthe problemtobe how the effecthe havingofwhich is thefunction f the heartbeat circulation) s to be dis-tinguishedfromothereffects f the heartbeat e.g., heartsounds).His answer s that circulation, ut notheartsounds,nsures neces-saryconditionfor the "properworkingof the organism."Thus,Hempel proposes 2) as an analysis f 1).(2) The heartbeat n vertebrates as the effect f circulating he blood,and this ensures the satisfaction f certain conditions (supply ofnutrimentnd removalofwaste)which are necessary orthe properworking fthe organism.

As Hempel seesthe matter, hemainproblemwiththis nalysis sthat functional tatementso construed ppear to have no explana-toryforce.Since he assumes A), theproblemforHempel is to seewhether 2) can be construed s a deductivenomologicalexplanansforthe presenceof theheartbeat n vertebrates,nd, in general,tosee whether tatements avingthe formof (2) can be construed sdeductive nomologicalexplananda forthepresence n a system fsome trait r itemthat s functionallyharacterized.


4/26

FUNCTIONAL ANALYSIS 743Suppose, hen,hatwe are nterestedn explainingheoccurrencef atrait in a system(at a certain ime ), and that he followingunc-tional nalysissoffered:(a) Att, functionsdequatelyna settingfkind (characterizedby pecificnternalndexternalonditions).(b) s functionsdequatelyn a settingfkindc only f a certainnecessaryondition,, ssatisfied.(c) f trait were resentns then,s an effect,ondition wouldbesatisfied.(d) Hence, t t, trait is presentns (Hempel, 10).

(d), of course,does not followfrom a)-(c), sincesometrait ' differ-ent from mightwell sufficeor the satisfactionf conditionn. Theargument an be patchedup bychanging c) to (c'): "Conditionnwouldbe satisfiedn s only f trait werepresentn s," but Hempelrightly ejectsthis avenue on the groundsthat nstances f the re-sulting chemawould typically e false. t is false,forexample,thattheheart s a necessaryonditionfor irculationn vertebrates,inceartificial umpscan be, and are,used tomaintaintheflow fblood.We are thus eftwith a dilemma. f theoriginalschema s correct,thenfunctional xplanation s invalid. f theschema s revised o asto ensure the validity of the explanation, the explanation willtypically e unsound,havinga falsethird remise.ErnestNagel offers defenseof what is substantiallyHempel'sschemawith c) replacedby c').

. . . a teleologicaltatementf theform,The functionfA in a sys-temS withorganization is to enable S in theenvironment toengage n process ," can be formulated oreexplicitlyy: everysystemwith rganizationand nenvironmentengagesnprocessP; ifS with rganization and in environmentdoesnothaveA,then doesnotengagenP; hence, with rganization musthaveA (Nagel, 03).

Thus he suggestshat 3) is to be rendered s (4):(3)The functionfchlorophylln plants s to enablethem operformphotosynthesis.(4)A necessaryonditionor he ccurrencefphotosynthesisnplantssthepresencefchlorophyll.So Nagelmustface the secondhornofHempel's dilemma: 3) is pre-sumably true,but (4) maywell be false.Nagel is, of course,awareof thisobjection.His rather uriousresponse s that,as far as weknow, hlorophyllsnecessaryorphotosynthesisn thegreenplants(404). This maybe so, but theresponsewill not survivea changeofexample.Heartsarenotnecessaryorcirculation,rtificial umps


5/26

744 THE JOURNAL OF PHILOSOPHYhaving actually been incorporated nto the circulatory ystems fvertebratesn such a way as to preserve irculation nd life.A more promisingdefense of Nagel might run as follows. Al-though it is true that the presence of a working heart is not anecessary ondition of circulation n vertebrates nder all circum-stances, till, under normal circumstances-most ircumstancesnfact-a working eart s necessary orcirculation. hus it is perhapstrue that,at the present tage of evolution, a vertebrate hat hasnot been tamperedwith urgicallywould exhibit circulation nly fit wereto containa heart. f thesecircumstancesre specificallyn-cluded in theexplanans, perhapswe can avoid Hempel's dilemma.Thus, insteadof 4) we should have:(4')At thepresent tage fevolution, necessaryonditionor ircula-tion nvertebrateshathavenotbeen urgicallyampered iths theoperation f a heart properlyncorporatednto the circulatorysystem).(4'), in conjunctionwithstatements sserting hata given vertebrateexhibits irculation nd has not been surgically amperedwith andis at the present tage of evolution,will logically mplythat thatvertebrate as a heart. t seems, hen, hat theHempelian objectioncould be overcomef t werepossible,givena truefunction-ascribingstatementike (1) or (3), tospecifynormalcircumstances"n such awayas tomake it truethat, n thosecircumstances,hepresenceofthe tem n question s a necessaryonditionfortheperformancefthefunction scribed oit.This defense as someplausibility s longas westick o theusualexamples drawn frombiology.But if we widen our view a bit,even withinbiology, think t can be shown thatthisdefenseofNagel's positionwill not suffice. onsider thekidneys. he functionof thekidneys s to eliminatewastesfrom he blood. In particular,thefunction fmy eftkidney s to eliminatewaste frommyblood.Yet thepresenceofmy eftkidney s not, n normalcircumstances,a necessaryonditionforthe removalof the relevantwastes.Only ifsomething eriously bnormalshouldbefallmy rightkidneywouldthe operationof my leftkidneybecome necessary, nd this onlyon theassumption hat am not hooked up to a kidneymachine."

1It mightbe objected here that,although it is the functionof the kidneystoeliminatewaste, that is not the functionof a particular kidneyunless operationof that kidney is necessaryfor removal of wastes. But suppose scientistshadinitially been aware of the existence of the left kidney only. Then, on theaccount being considered,anything theyhad said about the function of thatorgan would have been false, since, on that account, it has no function inorganismsaving wokidneys!


6/26

FUNCTIONAL ANALYSIS 745A less obvious counterexample erivesfrom he well-attestedactof hemispherical edundancyn the brain.No doubt it is in prin-ciple possible to specify onditionsunder which a particulardupli-cated mechanismwould be necessary or normal functioning fthe organism, ut (a) in mostcases we are not in a position actuallyto do this, houghwe are in a position to makewell-confirmedtate-ments about the functions f some of thesemechanisms, nd (b)these circumstancesre by no means the normalcircumstances.n-deed, given the fact that each individual nervoussystem evelopssomewhatdifferentlywing to differingnvironmental actors, hecircumstances n question might well be differentor each indi-

vidual, or for he same ndividualat differentimes.ApparentlyNagel was pursuingthe wrong trategyn attemptingto analyze functional scriptions n termsof necessary onditions.Indeed, we are still faced withthe dilemmanoticed by Hempel: ananalysis n terms f necessary onditionsyields valid but unsoundexplanatory chema; analysis n terms f sufficientonditions longthe lines proposedby Hempel yields a schemawithtruepremises,but validitys sacrified.Something as gone wrong, nd it is not too difficulto locate theproblem. An attempt to explain the presence of something byappeal to what it does-its function-is bound to leave unex-plained why somethingelse that does the same thing-a func-tional equivalent-isn't there nstead. n itself, his s not a seriousmatter.But the accountswe have been considering ssume thatexplanation is a species of deductiveinference, nd one cannotdeduce heartsfrom irculation.This is whatunderliesthedilemmawe have been considering.At best, one can deduce circulatorsfromcirculation. f we make this amendment,however,we areleftwitha functionallyainted analysis; the function f theheartis to circulatethe blood' is rendered a blood circulator s a (neces-sary/sufficient)onditionof circulation, nd the heartis a bloodcirculator'.The expression n italics is surelyas much in need ofanalysis as the analyzed expression.The problem,however,runsmuchdeeperthan thefact thattheperformancef a certainfunc-tion does not determinehow that function s performed.The

problem is rather that to "explain" the presenceof the heart invertebrates y appeal to what the heart does is to "explain" itspresencebyappeal to factors hat are causally rrelevant o itspres-ence.Even if twerepossible, s Nagel claimed,to deduce thepres-ence of chlorophyllfromthe occurrence of photosynthesis,hiswould fail to explain the presenceof chlorophyll n greenplantsin just the way deducing the presenceand heightof a building


7/26

746 THE JOURNAL OF PHIlLOSOPHYfromthe existence nd lengthof its shadowwould fail to explainwhythe building is thereand has the height t does. This is notbecause all explanationis causal explanation: it is not. But to ex-plain in thepresenceof a naturally ccurring tructure r physicalprocess-to explain why t is there,whysuch a thingexists n theplace (system,ontext) t does-this does requirespecifyingactorsthatcausallydetermine heappearanceof thatstructurer process.2There is, of course,a sense in which the question, "Why is xthere?" s answeredbygivingx's function.Considerthe followingexchange.X asks Y, "Why is that thing there (pointing to thegnomonof a sundial)?" Y answers, Because it casts a shadow onthe dial beneath, thereby ndicatingthe time of day." It is ex-changes of this sort that most philosophershave had in mindwhen theyspeak of functionalexplanation. But it seems to methat,althoughsuch exchangesdo represent enuine explanations,the use of functional anguage in this sortof explanationis quitedistinct rom ts explanatoryuse in science. n section II below, Iwill sketchwhat I thinkis the centralexplanatoryuse of func-tional language in science.Meanwhile,if I am right,the evidentpropriety f exchanges ike that imaginedbetweenX and Y hasled to prematureacceptance of (A), and hence to concentrationon what is, from the point of view of scientific xplanation, anirrelevant se of functional anguage. For it seemsto me that thequestion,"why s x there?" an be answeredby specifying 's func-tion only if x is or is part of an artifact.Y's answer, think,ex-plains thepresence f thegnomonbecause it rationalizes heactionoftheagentwho put it therebysupplyinghis reasonforputting tthere. n general,when we are dealing with the result of a de-liberate ction,we mayexplain theresultbyexplainingtheaction,and we may explain a deliberateaction by supplyingthe agent'sreasonfor t. Thus, whenwe look at a sundial,we assumewe knowin a generalwayhow the gnomoncame to be there: someonede-liberatelyput it there.But we maywish to know why t was putthere. Specifying he gnomon's function allows us to formulatewhatwe suppose to be the unknownagent'sreasonforputting tthere:he believed twould casta shadowsuchthat . . , and so on.

2 Even in the case of a designedartifact,t is at most the designer'sbeliefthatx will performf in s which is causally relevant to x's presence in s, not x'sactually performing in s. The nearest I can come to describinga situation inwhich x performingf in s is causally relevant to x's presence in s is this:the designerof s noticesa thing ike x performing in a system ike s, and thisleads to belief that x will perform in s, and this in turn leads the designertoput x in s.


8/26

FUNCTIONAL ANALYSIS 747When we do this,we are elaboratingon what we assume is thecrucial causal factor n determining he gnomon'spresence,namelya certaindeliberate ction.If this s on the righttrack, hen the viabilityof the sortof ex-planation in question should depend on the assumption that thethingfunctionally haracterizeds there as the result of deliberateaction. If that assumption s evidently alse, specifying he thing'sfunctionwill not answerthe question. Suppose it emerges hat thesundial is not, as such, an artifact.When the ancient buildingwasruined, a large stone fragment ell on a kind of zodiac mosaicand embedded itselfthere. Since no sign of the room remains,Yhas mistakenly upposedthe thingwas designedas a sundial.As ithappens, the local people have been using thething o tell timeforcenturies; o Y is right bout the function f thethingX pointedto.8But it is simply false that the thing is there because it casts ashadow, for there s no agent who put it there"because it castsashadow."Again, the function f a bowl-likedepression n a hugestonemaybe to hold holy water,but we cannotexplain why t isthereby appeal to itsfunction f we knowit was left thereby pre-historic lacial activity.If this s right, hen A) will lead us to focuson a type ofexpla-nation which will not apply to natural systems: hlorophyll ndhearts are not "there"as the result of any deliberate action, andhence the essentialpresupposition f the explanatorymove in ques-tion is missing.Once this becomes clear,to continueto insistthatthere must be some sense in which specifying he functionofchlorophyllexplains its presence is an act of desperation bornof thinking here s no otherexplanatory se of functional harac-terizationn science.Why have philosophers identifiedfunctionalexplanation ex-clusivelywith the appeal to something's unctionn explaining whyit is there?One reason, suspect, s a failure to distinguish eleo-logical explanation fromfunctionalexplanation,perhaps becausefunctional concepts do loom large in "explanations" having ateleologicalform.Someonewho fails to make this distinction, utwho senses that there s an important nd legitimateuse of func-

3 Is casting a shadow the function f this fragment? tandard use may conferfunction on something: if I standardly use a certain stone to sharpen knives,then that is its function,or if I standardly use a certain block of wood as adoor stop, then the function of that block is to hold my door open. If non-artifacts ever have functions, ppeals to those functionscannot explain theirpresence. The things functionally characterized in science are typically notartifacts.


9/26

748 THE JOURNAL OF PHILOSOPHYtional characterizationn scientificxplanation, will see the prob-lem as one of finding legitimate xplanatory ole forfunctionalcharacterizationwithinthe teleological form.Once we leave arti-facts nd go to naturalsystems, owever, his pproach is doomed tofailure, s critics fteleology ave seen for ome time.This mistakeprobablywould have sorted tself ut in time wereit not that we do reasonfrom he performance f a function o thepresence of certain specific processes and structures, .g., fromphotosynthesiso chlorophyll, r from oordinated ctivity o nervetissue.This is perfectlyegitimate easoning: t is a speciesof infer-ence to the bestexplanation. Our best (only)explanation of photo-synthesis equireschlorophyll, nd our best explanationof coordi-nated activity equires nerve tissue. But once we see what makesthis reasoning egitimate,we see immediatly hat inference o anexplanation has been mistaken or an explanationitself.Once thisbecomes clear, it becomes equally clear that (A) has mattersre-versed: given that photosynthesiss occurring n a particularplant,we may legitimatelynferthat chlorophyll s present n thatplantprecisely ecause chlorophyll nters nto our best only)explanationofphotosynthesis,nd givencoordinated ctivity n the partofsomeanimal, we maylegitimately nferthat nervetissue s presentpre-cisely because nervetissue enters nto our best explanationof co-ordinated ctivityn animals.To attempt o explain the heart's presence n vertebrates y ap-pealing to its function n vertebrates s to attempt o explain theoccurrence f hearts n vertebrates y appealing to factors hatarecausally rrelevant o itspresence n vertebrates.his facthas given"functional xplanation"a bad name. But it is (A) that deserves heblame. Once we see (A) as an undefendedphilosophicalhypothe-sis about how to construe unctional xplanationsratherthan as astatement f the philosophical problem,the correct lternative sobvious: what we can and do explain by appeal to whatsomethingdoes is the behaviorof a containing ystem.4A much morepromising uggestionn the ightof theseconsider-ations is that (1) is appealed to in explaining circulation. f wereject A) and adopt this uggestion, simple deductive-nomologicalexplanation with circulation s the explicandumturnsout to be asoundargument.

4 A confused perceptionof this factno doubt underlies (B), but the fact that(B) is nearly inseparable from (A) in the literature shows how confused thisperception s.


10/26


11/26

750 THE JOURNAL OF PHILOSOPHYThe secondsortof use occursas partof an accountwhich, f weare not careful, an easilybe mistakenfor an explanation of thepresenceof the sort of item functionally haracterized, nd thishas perhaps encouragedphilosophersto accept (A). For it mightseem thatnaturalselectionprovides hemissing ausal linkbetweenwhat somethingdoes in a certain typeof organismand its pres-ence in thattypeof organism.By performingheirrespective unc-tions,the contractile acuole and the neurofibrils elp species in-corporating hemto survive, nd thereby ontribute o theirowncontinuedpresence n organisms f those species,and thismightseem to explain the presenceof thosestructuresn the organismsincorporatinghem.Plausible as this sounds, it involves a subtle yet fundamentalmisunderstandingf evolutionary heory.A clue to the mistake sfound in the fact that the contractilevacuole occurs in marineprotozoans hat have no excess-waterroblembut thereverse rob-lem. Thus the function nd effect n survivalof this structure snotthe same n all protozoans. et the explanationof ts presence nmarine and fresh-waterpecies is almostcertainly he same. This

factremindsus that the processes ctuallyresponsiblefor the oc-currenceof contractilevacuoles in protozoansare totally nsensi-tive to what that structure oes. Failure to appreciate this pointnot only lends spuriousplausibility o (A) as applied to biologicalexamples; it seriouslydistorts ur understanding f evolutionarytheory.Whether n organismo incorporates dependson whethers is "specified"by thegenetic"plan" which o inherits nd which,at a certain evelofabstraction,s characteristicf o's species.Altera-tions n theplan are due tomutation. f a plan is alteredso that tspecifies ' ratherthan s, then the organisms nheriting his planwill incorporate t, regardless f the function r survivalvalue ofs' in thoseorganisms.f thealteration s advantageous, henumberof organismsnheriting hatplan mayincrease, nd, if it is disad-vantageous, heirnumbermaydecrease.But thishas no effect hat-everon theplan,and thereforeo effect hatever n theoccurrenceofst n theorganismsn question.One sometimes ears it said thatnatural selection s an instanceofnegativefeedback. f this s meantto implythat the relative uc-cess or failureof organisms f a certaintypecan affect heirin-heritedcharacteristics,t is simply mistake: the characteristicsforganismswhich determinetheir relative success are determinedby theirgeneticplan, and the characteristicsf these plans areutterly ndependentof the relative success of organismshaving


12/26

FUNCrIONAL ANALYSIS 751them.Of course, f s is verydisadvantageous o organismshavinga plan specifying , then organismshaving such plans may dis-appear altogether,nd s will no longeroccur.We could, therefore,thinkofnatural selection s reacting n the set of plans generatedby mutationby weedingout thebad plans: naturalselection annotalter a plan, but it can trimthe set.Thus, we may be able to ex-plain why a givenplan is not a failureby appeal to the functionsofthestructurest specifies. erhapsthis s what somewriters avehad in mind.But this s not to explainwhy, .g.,contractile acuolesoccur in certainprotozoans, t is to explain whythe sort of proto-zoan incorporatingontractile acuolesoccurs.Since we cannotap-peal to therelative uccessor failureof theseorganisms o explainwhytheirgeneticplan specifies ontractile acuoles, we cannotap-peal to therelative uccessor failureof theseorganisms o explainwhythey ncorporate ontractile acuoles.Once we are clear about the explanatoryrole of functions nevolutionary theory, t emergesthat the functionof an organor process or whatever) s appealed to to explain the biologicalcapacitiesof the organism ontaining t, and fromthese capacitiesconclusions re drawn concerning he chancesof survivalfor orga-nismsof thattype.For instance, ppeal to the function f the con-tractile acuole in certainprotozoans xplainshow theseorganismsare able to keep from xploding n freshwater.Thus evolutionarybiologydoes notprovide supportfor A) but for theidea instancedin (5): identifyinghe function f somethinghelps to explain thecapacities f a containingystem.6(A) misconstruesunctional xplanationbymisidentifyinghatisexplained. Let us abandon (A), then, in favorof the view thatfunctions re appealed to in explainingthe capacitiesof contain-ingsystems,nd turn urattention o (B).Whereas (A) is a thesisabout functionalexplanation, (B) is athesis bout the analysisof function-ascribingtatements. erhapswhendivorcedfrom A), as it is in (5), it will fare betterthan itdoes in the accountsof Hempel and Nagel.

6 In addition to the misunderstandingabout evolutionary theory ust dis-cussed,biological examples have probablysuggested A) because biologywas thelocus classicus of teleological explanation. This has perhaps encouraged a con-fusionbetweenthe teleologicalformofexplanation, ncorporated n (A), with theexplanatory role of functional ascriptions. Function-ascribingstatements dooccur in explanations having a teleological form,and, when they do, theirinterest s vitiated by the incoherenceof that formof explanation. It is thelegitimateuse of function-ascribingtatements hat needs examination, .e., theircontribution o nonteleologicaltheories uch as the theory f evolution.


13/26

752 THE JOURNAL OF PHILOSOPHYI1In spite of the evidentvirtues f (5), (5a) has serious hortcomingssan analysis f (1). In fact t is subjectto thesameobjectionHempelbringsto the analysis that simplyreplaces 'function'by 'effect':vertebratesncorporating workingheart n theusual wayexhibittheproduction f heartsounds, et theproduction fheartsoundssnot a functionof hearts in vertebrates.The problem is that,whereas the productionof certaineffectss essentialto the heart'sperforming ts function,there are some effectsproduction ofwhichis irrelevant o the functioning f the heart. This problemis bound to infectany "selected-effects"heory, .e., any theory

built on (B).What is needed to establisha selected-effectsheorys a generalformula that identifies he appropriateeffects.7oth Hempel andNagel attempt o solve this problemby identifyinghefunction fsomethingwith ust thoseeffectswhich contribute o the mainte-nance ofsomespecial conditionof,or theperformancef somespe-cial activity f, some containing ystem.f thissortof solution isto be viable, theremust be some principledway of selectingtherelevant ctivities r conditionsof containing ystems. or no mat-ter whicheffectsf somethingyou happen to name, therewill besome activity fthe containing ystemo which ust thoseeffectson-tribute, r some conditionof thecontaining ystemwhich s main-tained with the help of just those effects. eart activity, or ex-ample,keeps the circulatory ystem rombeingentirely uiet, andthe appendixkeepspeople vulnerableto appendicitis.8Hempel suggests hat, n general, hecrucial feature fa contain-ing system,ontributionowhich s to count as thefunctioningfa

7 Larry Wright op. cit.) is aware of this problem,but does not, to my mind,make much progresswith it. Wright'sanalysis rules out "The functionof theheart is to produce heart sounds," on the ground that the heart is not therebecause it produces heartsounds. I agree. But neither is it there because itpumps blood. Or if,as Wrightmaintains,there is a sense of 'because' in whichthe heart is there because it pumps blood and not because it produces heart-sounds, then this sense of 'because' is as much in need of analysis as 'func-tion'. Wrightdoes not attemptto provide such an analysis,but depends on thefact that, in many cases, we are able to use the word in the required way.But we are also able to use 'function' correctlyn a variety of cases. Indeed, ifWrightis right,the words are simply nterchangeablewith a little grammaticalmaneuvering.The problem is to make the conditions of correct use explicit.Failure to do thismeans thatWright's nalysisprovidesno insight nto the prob-lem of how functional theories are confirmed, r whence they derive theirexplanatoryforce.8 Surprisingly,when Nagel comes to formulatehis general schema of func-tional attributionhe simply gnores this problem,and thus leaves himselfopento the trivialization ust suggested.Cf.,Nagel, op. cit., p. 403.


14/26

FUNCrIONAL ANALYSIS 753contained part, is that the system e maintained n "adequate, oreffective, r proper working order" (306). Hempel explicitlyde-clines to discuss what constitutes proper working order," pre-sumablybecause he rightly hinks hat there re moreseriousprob-lems with the analysis he is discussing than those introducedbythis phrase.But it seemsclear that for something o be in workingorder s justfor t to be capable ofperformingts functions, nd forit to be in adequate or effectiver properworking rder s just forit to be capable of performingtsfunctions dequately or effectivelyorproperly.Hempel seemsto realizethishimself, or n setting ortha deductive chemaforfunctional xplanation,he glosses hephrasein question as 'functions dequately' (310). More generally, f weidentify he function f something with those effectsf x whichcontribute o the performance f some activity or to themainte-nance of some condition c of a containing ystem , then we mustbe preparedto say as well that a function f s is to perform or tomaintain c. This suggests he followingformulation f "selected-effects"heories.(6)The functionfan F ina G isf ust ncase the apacity or) s aneffectf an F incorporatedn a G in the usual way or: in thewaythisF is incorporatedn thisG), and that ffectontributeso theperformancef functionf thecontaining.It seems that any theorybased on (B)-what I have been calling"selected-effects"heories-must ultimatelyamount to somethinglike (6).9 Yet (6) cannot be the whole story about functionalascriptions.Supposewe follow 6) in rendering,The function fthecontrac-tile vacuole in protozoans s elimination of excesswater from theorganism." he result s (7):(7)Elimination fexcesswater romheorganisms an effectfa con-tractileacuole ncorporatednthe sualwayna protozoan,ndthateffectontributeso theperformancefa functionf a protozoan.In orderto test 7) we should have toknow a statementftheform,"f is a function f a protozoan."Perhaps protozoanshave no func-tions. f not, (7) is just a mistake. f theydo, thenpresumablyweshall have to appeal to (6) for an analysisof thestatementttribut-ing such a function, nd thiswill leave us withanotherunanalyzedfunctional scription.Eitherwe are launched on a regress, r the

9Hugh Lehman (op. cit.) given an analysis hat appearsto be essentiallylike 6).


15/26

754 THE JOURNAL OF PHILOSOPHYanalysisbreaksdown at some evel for ack of functions, r perhapsfor ack of a plausible condidatefor ontaining ystems.f we do notwish simply o acquiesce in theautonomy f functional scriptions,it must be possible to analyze at least some functional scriptionswithout ppealing to functions f containing ystems.f (6) can beshown to be the only plausible formulation f theoriesbased on(B), then no such theory an be the whole story.Our question, hen, s whether thing's unction an plausiblybeidentifiedwith those of its effects ontributing o productionofsomeactivity f, or maintenance f some conditionof, a containingsystem,where performance f the activity n question is not afunction of the containing system.Let us begin by consideringHempel's suggestion hat functions re to be identifiedwith pro-duction of effects ontributingto proper working order of acontaining ystem. claimedearlier on that to say somethings inproper workingorder is just to say that it properlyperforms tsfunctions.This is fairlyobvious in cases of artifacts r tools. Tomake a decision about which sort of behavior counts as workingamounts to deciding about the thing'sfunction.To saysomethingis working, hough not behavingor disposed to behave in a wayhaving anythingto do with its function, s to be open, at thevery east, o the changeofarbitrariness.When we are dealing with a living organism,or a societyof living organisms, hesituation s less clear.If we say,"The func-tionof the contractile acuole in protozoans s elimination fexcesswater fromthe organism,"we do make reference o a contain-ing organism,but not, apparently, o its function if any). How-ever, incecontractile acuolesdo a number fthings avingnothingto do with theirfunction, heremustbe someimplicitprincipleofselectionat work.Hempel's suggestions that, n this context, obe in "properworking rder" s simply o be alive and healthy.Thisworks reasonablywell for certainstandardexamples, e.g., (1) and(3): circulationdoes contributeto health and survival in verte-brates, nd photosynthesisoes contribute o health and survival ngreenplants.10 ut,onceagain,theprinciplewill notstand a changeof example,even withinthe life sciences.First,thereare cases inwhich proper functionings actuallyinimical to health and life:

10 Even these applications have their problems. Frankfurt nd Poole ["Func-tional Explanations in Biology,"BritishJournal for the Philosophy of Science,xvii (1966)] point out that heartsounds contribute to health and survival viatheirusefulnessn diagnosis.


16/26

FUNCTIONALI ANALYSIS 755functioning f thesex organsresults n the death of individuals ofmany species (e.g., certain almon). Second, a certainprocess n anorganism may have effectswhich contributeto health and sur-vival but whichare not to be confusedwith the function f thatprocess: secretion f adrenalin speeds metabolism nd thereby on-tributes o elimination of harmfulfat deposits in overweight u-mans,but this s not a function f adrenalin secretionn overweighthumans.A moreplausible suggestion long these ines in the special con-text fevolutionary iology s this:(8) The functionsf a part rprocessnan organismre to be identifiedwith hose f tseffectsontributingoactivitiesrconditionsf theorganism hich ustain r increase he organism'sapacity o con-tributeo survivalfthe pecies.Give or take a nicety, 8) doubtlessdoes capturea great many usesof functional anguage in biology.For instance, t correctly icksout eliminationof excesswater as the function f the contractilevacuole in fresh-waterrotozoansonly,and correctlydentifies hefunctionof sex organs in species in which the exerciseof theseorgansresults n the deathofthe ndividual."-In spite of these virtues, however, 8) is seriously misleadingand extremelyimited in applicability even within biology. Evi-dently,what contributes o an organism's apacityto maintain itsspecies in one sortof environmentmay undermine that capacityin another. When this happens,we might say that the organ (orwhatever) as lost tsfunction. his is probablywhatwe would sayabout the contractile acuole if fresh-waterrotozoansweresuccess-fully ntroduced nto salt water,for in this case the capacity ex-plainedwouldno longerbe exercised.But ifthecapacity xplainedby appeal to the functionof a certainstructure ontinuedto beexercised n thenew environment, houghnow to the individual'sdetriment,we would not say that thatstructure ad lost its func-tion. f, for somereason,flyingeased to contribute o thecapacityof pigeons to maintain their species, or even undermined thatcapacityto someextent,'2we would stillsaythat a function f thewings n pigeons s to enable themtofly.Only fthewingsceased to11 Michael Ruse has argued for a formulation ike (8). See "Function State-ments in Biology," Philosophy of Science, xxxviii, 1 (March 1971): 87-95, andThe Philosophy of Biology (London: Hutchinson, 1973).12 Perhaps, in the absence of serious predators,with a readily available foodsupply, and with no need to migrate,flying implywastesenergy.


17/26

756 THkE JOURNAL OF PHILOSOPHYfunction s wings, s in the penguinsor ostriches,would we ceaseto functionallynalyzeskeletalstructurend the like with an eyeto explainingflight. light s a capacitythat criesout forexplana-tionin terms f anatomicalfunctions egardless f its contributionto thecapacity o maintainthespecies.What thisexampleshows s that functional nalysiscan properlybe carriedon in biology quite independently f evolutionary on-siderations: complexcapacityof an organism or one of itspartsor systems)may be explained by appeal to a functional nalysisre-gardlessof how it relates o theorganism's apacityto maintainthespecies.At best,then, 8) picksout thoseeffects hich will be calledfunctionswhen whatis in theoffings an applicationof evolution-arytheory.As we shall see in the nextsection, 8) is misleading swell in that it is not which effectsre explained but the styleofexplanation that makes it appropriate to speak of functions. 8)simply dentifiesffects hich, s it happens,are typicallyxplainedin that tyle.We have not quite exhaustedthe lessons to be learnedfrom 8).The plausibility f 8) rests n the plausibility f the claim that,forcertainpurposes,we mayassume that a function f an organism sto contribute o the survival f ts species.What (8) does, n effect,sidentify function fan important lassof uncontained) ontainingsystems ithout roviding n analysis ftheclaimthata function fan organisms to contribute othe survival f ts species.Of course,an advocate of (8) might nsistthat it is no part ofhis theory o claim that maintenance f thespecies s a function fan organism.But then thedefenseof (8) would have to be simplythat t describes ctual usage, .e., that t is in facteffectsontribut-ing to an organism's apacityto maintainits specieswhichevolu-tionary iologists ingleout as functions. onstrued n thisway, 8)would, at most,tell us which effectsre pickedout as functions; twould provide no hint as to whythese effectsre picked out asfunctions.We know whyevolutionary iologistsare interestedneffectsontributingoan organism's apacity omaintain tsspecies,but whycall themfunctions?This is preciselythe sort of ques-tion that a philosophicalaccount of function-ascribingtatementsshouldanswer.Either 8) is defended s an instanceof (6)-mainte-nance of the species is declared a function f organisms-or it isdefended s descriptive fusage. n neither ase is any philosophicalanalysisprovided.For in the first ase (8) relieson an unanalyzed(and undefended) unction-ascribingtatement,nd in the second t


18/26

FUNCTIONAL ANALYSIS 757fails ogiveany hint s to the pointof identifyingertain ffectsasfunctions.

The failings f 8) are, think, oundto cripple nytheoryhatidentifies thing's unctions ith ffectsontributingo some nte-cedently pecified ypeof condition r behavior f a containingsystem.f the theorys an instance f (6), it launches regress rterminatesn an unanalyzedunctionalscription;f it is not aninstance f 6), then t isbound to leave open thevery uestion tissue, iz.,why re the electedffectseenas functions?IIIIn this ection, will ketchrieflyn account ffunctionalxplana-

tion which akes eriouslyhe ntuition hat t is a genuinely is-tinctivetyle f explanation.he assumptionsA) and (B) form hecore of approaches hat eekto minimize he differencesetweenfunctionalxplanationsnd explanations otformulatedn func-tional erms.uch pproaches avenotgivenmuch ttentiono thecharacterizationfthe pecial xplanatorytrategycience mploysin usingfunctionalanguage, ortheproblem s it was conceivedin suchapproaches as toshow hatfunctionalxplanations notreally ifferentn essentialsrom therkinds f scientificxplana-tion. Once the problem s concenvedn thisway,one is almostcertain o missthe distinctiveeaturesffunctionalxplanation,andhence omiss hepoint ffunctionalescription.he accountofthis ection everseshis endencyyplacing rimarymphasis nthekindof problem hat s solvedbyappeal to functions.1.Functions nd Dispositions.Something arbe capableofpump-ing even though t does not functions a pump ever) nd eventhough umpingsnot tsfunction.n the ther and,f omethingfunctionss a pump n a systemor ifthefunction fsomethingin a systemis to pump, hent must e capableofpumpingn s."s

13Throughout his ection am discountingppealsto the ntentionsfde-signers r users. maybe intended o preventccidents ithoutctually eingcapableof doingso. Withreferenceo this ntentiont wouldbe proper ncertain ontexts o say, x's functions to prevent ccidents, hough t is notactually apableofdoing o."There can be no doubtthata thing's unctions oftendentified ithwhatit is typicallyr "standardly"sed to do, or withwhat t was designed o do.But thesorts f things orwhich t is an importantcientificroblem o pro-vide functionalnalyses-brains, rganisms,ocieties,ocial institutions-eitherdo nothave designers r standard r regularuses at all, or it would be in-appropriate o appeal to these n constructingnd defining scientificheorybecause hedesignerr use is notknown-brains,evices ugup byarchaeolo-


19/26

758 THE JOURNAL OF PHILOSOPHYThus, function-ascribingtatementsmplydispositionstatements;to attributea functionto something s, in part, to attributeadisposition o it. If the function fx in s to p,thenx has a disposi-tion to d in s. For instance,f thefunction f the contractile acuolein fresh-waterrotozoans is to eliminate excess water from theorganism, hen there mustbe circumstances nderwhich the con-tractilevacuole would actuallymanifest dispositionto eliminateexcesswaterfrom he protozoan hat ncorporatest.To attribute dispositiond to an object a is to assert that thebehavior of a is subject to (exhibitsor would exhibit) a certainlawlikeregularity: o say a has d is to say that a would manifestd (shatter, issolve)wereany of a certainrange of events to occur(a is put in water, a is struck harply).The regularity ssociatedwith a disposition-call it the dispositionalregularity-is a regu-larity hat s special to the behavior ofa certainkindof object andobtainsin virtueof some special fact(s)about thatkind of object.Not everythings water-soluble:uchthings ehave in a specialwayin virtue of certain (structural)features pecial to water-solublethings.Thus it is thatdispositions equire explanation: ifx has d,thenx is subject to a regularityn behavior pecial to thingshavingd, and such a fact needs to be explained.To explain a dispositionalregularitys to explain how mani-festations f the dispositionare brought bout given the requisiteprecipitatingonditions.n whatfollows, will describe wodistinctstrategies or ccomplishinghis. t is mycontention hat theappro-priatenessof function-ascribingtatements orresponds o the ap-propriateness f the-second f these two strategies. his, I think,explainstheintuition hat functional xplanation s a special kindofexplanation.2. Two Explanatory trategies(i) The Subsumption trategy. upposea has a dispositiond. Theassociateddispositionalregularity onsists n the fact that certainkindsof eventswould cause a to manifest . One way to explainthisfactwouldbe todiscover ome feature fa whichallowedus torepresent he connectionbetweenprecipitating vents and mani-gists-or because there is some likelihood that real and intended functiondiverge-social institutions, omplex computers.Functional talk may have origi-nated in contexts n which reference o intentionsand purposes loomed large,but referenceto intentionsand purposes does not figureat all in the sort offunctional nalysisfavoredby contemporary atural scientists.


20/26

FUNCTIONAL ANALYSIS 759festations s instances f one or more general aws, .e., laws govern-ing the behaviorof things enerally, ot ust thingshavingd. BrianO'Shaughnessy as provided n example whichallowsa particularlysimple illustrationof this strategy.'4 onsider the dispositionhecalls elevancy:the tendency f an object torise in water of its ownaccord. To explain elevancy,we must explain why freeing sub-merged levant object causes t to rise.15 his we may do as follows.In every case, the ratio of an elevant object's mass to its non-permeablevolume s less than the density massper unit volume)ofwater. Archimedes'principle tells us thatwater exerts an upwardforceon a submerged bject equal to the weightof the water dis-placed. In the case of an elevantobject, thisforce vidently xceedsthe weightof the object by some amount f.Freeing the objectchanges thenet forceon it from ero to a net forceofmagnitudef in the directionof the surface, nd the object rises accordingly.Here, we subsumethe connectionbetweenfreeingsnd risingsun-der a general aw connecting hanges n net forcewithchanges nmotion by citing a featureof elevant objectswhichallows us (viaArchimedes' rinciple) to represent reeing hemunderwater as aninstance f introducing net force n thedirection fthe surface.(ii) The Analytical Strategy.Rather than subsume a disposi-tional regularity nder a law not special to the disposed objects,the analytical strategy roceedsby analyzinga disposition d of ainto a number of other dispositionsd1 . . . dn, ad by a or compo-nentsof a such that programmedmanifestationf the di results nor amountsto a manifestation fd.16The twostrategieswill fit o-gether nto a unifiedaccount if the analyzingdispositions the di)can be made toyieldtothesubsumptiontrategy.When theanalyticalstrategys in theoffing,ne is apt to speakof capacities (or abilities) rather than of dispositions.This shiftin terminologywill put a more familiarface on the analytical

14 "The Powerlessnessof Dispositions," Analysis, xxxi.1, 139 (October 1970):1-15. See also mydiscussionof thisexample in "Dispositions,States and Causes,"ibid., xxxiv.6, 162 (June 1974): 194-204.15Also, we must explain why submerginga free elevant object causes it torise, and why a free submerged object's becoming elevant causes it to rise. Oneof the convenientfeaturesof elevancy s that the same considerationsdispose ofall thesecases. This does not hold generally: gentle rubbing,a sharp blow, or asudden change in temperaturemay each cause a glass to manifest dispositionto shatter,but the explanations in these cases are significantlyifferent.

16 By "programmed" simply mean organizedin a way that could be specifiedin a program or flow chart: each instruction box) specifiesmanifestationofone of the d4 such that, if the program is executed (the chart followed), amanifests .


21/26

760 THE JOURNAL OF PHILOSOPHYstrategy,17orwe oftenexplain capacitiesby analyzingthem.As-sembly-line roductionprovidesa transparent xample of what Imean. Production s brokendown into a numberof distinct asks.Each pointon the ine is responsible or certaintask, nd it is thefunctionof the workers/machinest thatpoint to completethattask. f the inehas thecapacity o producetheproduct, t has it invirtueof thefactthattheworkers/machinesave the capacitiestoperform heirdesignated asks, nd in virtueof thefact thatwhenthese asks re performedn a certain rganizedway-according to acertainprogram-the finished roductresults.Here we can explainthe line's capacityto produce the product-i.e., explain how it isable to producetheproduct-by appeal to certaincapacitiesof theworkers/machinesnd their organizationinto an assembly ine.Againstthisbackground,we maypick out a certaincapacityof anindividualexercise fwhich s hisfunction n the ine.Of themanythingshe does and can do, his function n the line is doingwhat-ever t is thatwe appeal to in explainingthe capacityof the ine asa whole. If the line produces everalproducts, .e., if it has severalcapacities,then,althougha certaincapacityc of a worker s irrele-vant to one capacityof the line, exerciseof c by that workermaybe his functionwith respectto anothercapacityof the line as awhole.Schematicdiagrams n electronics rovideanotherobvious illus-tration.Since each symbolrepresentsny physicalobject whateverhavinga certaincapacity, schematic iagramof a complexdeviceconstitutesn analysisof theelectronic apacitiesof the device as awhole intothecapacitiesof its components. uch an analysis llowsus to explain how thedevice as a whole exercises he analyzedca-pacity,for t allows us to see exercisesof the analyzedcapacityasprogrammed xerciseof the analyzingcapacities. n this case, the"'program"s givenbythelines indicatinghow thecomponents rehooked up. (Of course,the lines are themselves unction ymbols.)Functional analysis in biology is essentially imilar. The bio-logicallysignificantapacitiesof an entire organism re explained

17Somemightwantto distinguishetween ispositionsnd capacities,ndarguethat to ascribe functiono x is in partto ascribe capacity to x, nota dispositions I have claimed. ertainly1) is strainedn a way 2) is not.(1)Hearts redisposed opump.Heartshave dispositiono pump.Sugar scapableofdissolving.Sugarhas a capacityodissolve.(2)Hearts recapableofpumping.Heartshave capacityopump.Sugar sdisposed odissolve.Sugarhasa dispositionodissolve.


22/26

FUNCTIONAL ANALYSIS 76Iby analyzingtheorganism nto a numberof "systems"-thecircu-latory ystem,hedigestiveystem,he nervous ystem,tc.,-each ofwhichhas itscharacteristicapacities.1S hese capacitiesare in turnanalyzed ntocapacitiesofcomponent rgans nd structures.deally,thisstrategys pressed until pure physiology akes over, .e., untilthe analyzingcapacitiesare amenable to the subsumption trategy.We can easily maginebiologists xpressing heir nalyses n a formanalogous to the schematic iagramsof electrical ngineering, ithspecialsymbols orpumps,pipes, filters,nd so on. Indeed, analysesof even simple cognitive apacitiesare typically xpressed n flowcharts r programs, orms esigned pecificallyo represent nalysesof nformationrocessingapacitiesgenerally.Perhapsthemostextensive se ofthe analytical trategyn scienceoccurs n psychology,ora largepartof thepsychologist'sob is toexplain how the complex behavioral capacities of organismsareacquired and how they re exercised.Both goals are greatlyfacili-tatedby analysis f thecapacities n question,forthen cquisitionofthe analyzedcapacity resolves tself nto acquisitionof the analyz-ing capacities and the requisiteorganization, nd the problemofperformance esolves tself nto the problemof how the analyzingcapacitiesare exercised.This sort of strategy as dominatedpsy-chologyoversinceWatson attempted o explain such complexca-pacitiesas the ability to run a maze by analyzingthe performanceinto a series of conditionedresponses, he stimulusfor each re-sponsebeingthepreviousresponse r somethingncountered s theresult f theprevious esponse.19cquisition f thecomplexcapacityis resolved nto a numberof distinct ases of simple conditioning,i.e., the abilityto learn the maze is resolved nto the capacityforstimulus ubstitution,nd the capacityto run themaze is resolvedinto abilities to respondin certainsimpleways to certainsimplestimuli.Watson's analysis proved to be of limitedvalue, but theanalytic strategy emains the dominant mode of explanation inbehavioralpsychology.20

18 Indeed,whatmakes omethingartof, e.g.,the nervousystems that tscapacities iguren an analysis f thecapacity o respond o external timuli,coordinatemovement,tc.Thus,theres no question hat heglialcellsarepartof the brain,but there s somequestion s to whether hey re partof thenervousystemr merelyuxiliaryo t.19JohnB. Watson, ehaviorismNewYork:Norton, 930)chs. x andxi.20 Writers n the philosophy f psychology,speciallyJerry odor,havegrasped heconnection etween unctionalharacterizationnd the analyticalstrategyn psychologicalheorizing,ut have not applied the lessonto theproblem f functionalxplanation enerally. he clearest tatementccurs nJ.A. Fodor, The Appealto Tacit KnowledgenPsychologicalxplanation,"hisjouRNAL, Lxv, 24 (December 9,1968):627-640.


23/26

762 THE JOURNAL OF PHILOSOPHY3. Functionsand FunctionalAnalysis. n the contextof an appli-cation of the analyticalstrategy, xerciseof an analyzingcapacityemerges s a function: twill be appropriate o say thatx functionsas a pin s, or that thefunction f x in s is (-ing,whenwe are speak-ing against the backgroundof an analyticalexplanation of somecapacityof s whichappeals to the factthatx has a capacityto 4 ins. It is appropriate o say thatthe heartfunctionss a pump againstthe background f an analysis f the circulatoryystem's apacitytotransport ood,oxygen,wastes, nd so on, whichappeals to thefactthat the heart s capable of pumping.Since this s theusual back-ground, t goeswithout aying, nd thisaccounts forthe factthat"The heartfunctionss a pump" soundsright, nd "The heartfunc-tionsas a noise-maker" oundswrong, n some context-freeense.This effects strengthenedythe absenceof anyactual applicationofthe analytical trategy hichmakesuse of thefactthattheheartmakesnoise.2'We can capturethis mplicitdependenceon an analytical ontextby entering n explicit relativizationn our regimented econstruc-tion offunction-ascribingtatements:(9)x functionssa q in s (or: thefunctionfx in s is to0) relative oananalyticalccountA ofs's capacityo 'p ust n case x is capableof0-ingn s and A appropriatelynd adequatelyccounts or 's ca-pacity o pby, n part, ppealing othecapacityf x to p in s.Sometimeswe explain a capacityof s by analyzing t into othercapacities of s, as when we explain how someone ignorantofcookery s able to bake cakes by pointingout thathe followedarecipe each instructionf whichrequiresno specialcapacitiesfor tsexecution.Here, we don't speakof,e.g., stirrings a function f thecook,but ratherof the functionof stirring. ince stirring as dif-ferent unctionsn different ecipesand at differentoints in thesame recipe,a statementike 'The function f stirringhemixtureis to keep it from ticking o the bottomof the pot' is implicitlyrelativized o a certain perhapssomewhatvague) recipe. To takeaccount of this sortof case, we need a slightlydifferentchema:where e is an activity r behavior of a system (as a whole), thefunction f e in s is to (prelativeto an analyticalaccount A of s'scapacityto , just in case A appropriatelynd adequatelyaccountsfor 's capacityto q by, n part,appealing to s's capacityto engagein e.

21 It issometimesuggestedhatheartsoundso have a psychologicalunction.In the context f an analysis f a psychologicalisposition ppealingto theheart'snoise-makingapacity,The heartfunctionss a noise-maker"e.g.,asa producer f regular humps), ouldnoteven sound odd.


24/26

FUNCTIONAL ANALYSIS 763(9) explains the ntuitionbehind the regress-ridden6): functionalascriptions o requirerelativization o a "functionalfact" about acontaining ystem,.e., to the factthat a certain capacityof a con-taining systems approximately xplained by appeal to a certainfunctional nalysis.And, like (6), (9) makes no provisionfor speak-ing of the functionof an organismexcept against a backgroundanalysisof a containing ystem the hive, the corporation, he eco-system).Once we see that functions re appealed to in explain-ing the capacitiesof containing ystems,nd indeed that it is theapplicability of a certain strategy or explaining these capacitiesthat makes talk of functions ppropriate,we see immediatelywhy

we do not speak of the functions f uncontainedcontainers.What(6) fails to captureis the fact that uncontainedcontainers an befunctionallynalyzed, nd the wayin which function-analyticalx-planation mediatesthe connectionbetween functional scriptions(x functions s a 0, the functionof x is to q) and the capacitiesofthe containers.4. Function-analytical xplanation. If the account I have beensketching s to draw any distinctions,he availabilityand appro-priateness f analyticalexplanationsmust be a nontrivialmatter.22So let us examinean obviouslytrivial pplication of the analyticalstrategy ith an eyeto determining hether t can be dismissed nprincipledgrounds.(10)Eachpartofthemammalianirculatoryystem akes tsowndis-tinctiveound, ndmakes t continuously.hese combine o formthe "circulatoryoise" characteristicf all mammals. he mam-malian irculatoryystems capableof producinghis ound t vari-ous volumesnd various empos. he heartbeats responsibleorthethrobbingharacterfthesound, nd it is thecapacityftheheart obeatat various ates hat xplains hecapacityfthecircu-latory ystemo produce variouslyempoedound.Everythingn (10) is, presumably,rue. The questionis whether tallows us to saythat thefunction f theheart s to producea vari-ously tempoedthrobbing ound.28 o answer hisquestionwe must,I think, et clear about themotivationforapplyingthe analytical

22 Of course, it might be thatthere re none but arbitrarydistinctions o bedrawn. Perhaps (9) describes usage, and usage is arbitrary, ut I am unable totake thispossibility eriously.23The issue is not whether (10) forces us, via (9), to say somethingfalse.Relative to some analytical explanation, it may be true that the functionof theheart is to produce a variously tempoed throbbing.But the availability of (10)should not supportsuch a claim.


25/26

764 THE JOURNAL OF PHILOSOPHYstrategy. or my contentionwill be that the analytical strategy smost ignificantlypplied in casesveryunlike thatenvisaged n (10).The explanatory nterestof an analytical account is roughlyproportional o (i) the extent to which the analyzing apacities areless.sophisticatedthan the analyzed capacities, (ii) the extent towhich the analyzing capacities are differentn type from theanalyzed capacities, nd (iii) the relative ophistication f the pro-gramappealed to, .e., the relative omplexity f the organization fcomponentparts/processeshat is attributed o the system.iii) iscorrelativewith (i) and (ii): the greaterthe gap in sophisticationand type between analyzing capacities and analyzed capacities,themore sophisticated he programmust be to close thegap.It is precisely he width of thesegaps which,for nstance,makesautomata theory o interesting n its application to psychology.Automata theory upplies us with extremely owerfultechniquesfor constructing iverse analysesof very sophisticatedtasks intoveryunsophisticated asks.This allows us to see how, in principle,a mechanism uch as the brain, consisting f physiologically n-sophisticated components relatively peaking), can acquire verysophisticatedcapacities. It is the prospectof promotingthe ca-pacity to store ones and zeros into the capacity to solve logicproblems nd recognizepatterns hat makesthe analytical trategyso appealing in cognitivepsychology.As the program absorbs more and more of the explanatoryburden, the physical factsunderlying he analyzingcapacities be-come less and less special to the analyzed system. his is why t isplausible to suppose that the capacityofa person and of a machineto solve a certain problem mighthave substantially he same ex-planation,although t is notplausible tosuppose that thecapacitiesof a synthesizer nd of a bell to make similar sounds have sub-stantially imilar explanations. There is no work to be done bya sophisticatedhypothesis bout the organizationof various ca-pacities n the case of thebell. Conversely,helessNveightornebythe program, helesspoint to analysis.At thisend of the scale wehave cases like (10) in whichthe analyzedand analyzingcapacitiesdifferittle fat all in type nd sophistication. ere we could applythe subsumptionstrategywithoutsignificant oss, and thus talkof functionss comparativelytrained nd pointless. t mustbe ad-mitted,however, hatthere s no black-white istinctionhere,buta case of more-or-less.s the role oforganizationbecomes ess andless significant,he analytical strategy ecomes less and less ap-propriate, nd talkoffunctionsmakes essand lesssense.This maybe philosophically isappointing, ut there s no help for t.


26/26

NOOLREvEws 765CONCLUSIONAlmostwithout xception, hilosophical ccounts ffunction-ascrib-ing tatementsnd of functionalxplanation avebeencrippled yadoption f the assumptionsA) and (B). Though therehas beenwide-spreadgreementhat xtant ccounts re not satisfactory,A)and (B) have escaped ritical crutiny,erhaps ecausetheywerethought f as somehow etting he problem ather hen as partof profferedolutions.Once the problem s properly iagnosed,however,t becomes ossible o givea more atisfactorynd moreilluminatingccount n terms f theexplanatorytrategyhatpro-vides the motivationnd forms he context f function-ascribing

statements,o ascribe functiono somethings to ascibe a ca-pacity to it which is singled out by its role in an analysisof somecapacity f a containingystem.Whena capacity f a containingsystems appropriatelyxplained y analyzingt intoa number fother apacitieswhoseprogrammedxercise ields manifestationof the analyzed apacity,heanalyzing apacities merge s func-tions. incetheappropriatenessfthis ort fexplanatorytrategyis a matter f degree,o is the ppropriatenessffunction-ascribingstatements. ROBERT CUMMINSTheJohns opkinsUniversity

BOOK REVIEWSThe Anatomy f theSoul. ANTHONY KENNY. Oxford:Basil Black-well; NewYork:Barnes&Noble, 1973. x,147p. $8.50.Not ongbefore wasasked oreview his ook ordered copy fit from hecatalogue f itsBritish ublishers,lackwell, n theassumptionhata newbook of essayswouldnot,unlessclearlyindicated n catalogue nd advertisements,onsist f unrevisedreprintsf already ublishedmaterial. utno: notonlyhave allsix of these ssays ppearedpreviously,ut no fewer han fourof them re on myshelves n other ollections,hree n cheappaperbacks. heremustbe very ewteachersf philosophy,ndeven fewer ollegeand universityibraries,hatdo not alreadyownmost,fnotall,of thesematerials.t is natural o ask,then,what pecialvaluethisbookcanhavefor ts ntended eadersndbuyers.Well,theres somenewworkhere, n theform fan eighteen-pageAppendix, onsistingf an intendedncyclopediarticle n

Cummins- Functional Analysis

Documents