The Journal of Philosophy Volume 72 Issue 20 1975 (Doi 10.2307/2024640) Cummins, Robert - Functional Analysis

Journal of Philosophy, Inc.
Functional Analysis
Author(s): Robert Cummins
Source: The Journal of Philosophy, Vol. 72, No. 20 (Nov. 20, 1975), pp. 741-765
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2024640
Accessed: 22-06-2015 21:33 UTC
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 131.215.225.9 on Mon, 22 Jun 2015 21:33:04 UTC
All use subject to JSTOR Terms and Conditions
' +44
THE JOURNALOF PHILOSOPHY

VOLUME LXXII, NO 20, NOVEMBER 20, I975
FUNCTIONAL ANALYSIS
A on the na-
SURVEY of the recentphilosophicalliterature
ture of functionalanalysisand explanation,beginningwith
the classic essays of Hempel in 1959 and Nagel in 1961,
reveals that philosophicalresearchon this topic has almostwithout
exceptionproceededunderthefollowingassumptions:*
(A) The pointof functional in scienceis to explainthe
characterization
presenceof the item(organ,mechanism, processor whatever)that
characterized.
is functionally
(B) For somethingto performitsfunctionis forit to havecertaineffects
on a containingsystem,whicheffects
contribute to the performance
of someactivityof,or the maintenance of some conditionin, that
containingsystem.
Puttingthese two assumptionstogetherwe have: a function-ascrib-
ing statementexplains the presenceof thefunctionallycharacterized
item i in a systems by pointingout that i is presentin s because it
has certaineffectson s. Give or take a nicety,thisfusionof (A) and
(B) constitutesthe core of almost everyrecentattemptto give an
account of functionalanalysisand explanation. Yet these assump-
* Cf, Carl Hempel,"The Logic ofFunctionalAnalysis," in AspectsofScientific
Explanation(New York:Free Press,1965),reprintedfromLlewellynGross,ed.,
Symposiumon SociologicalTheory (New York: Harper & Row, 1959), and
ErnestNagel, The Structure of Science(New York: HarcourtBraceJovanovich,
1961),ch. 12, sec. x. The assumptions, of course,predateHempel's 1959 essay.
See, for instance,Richard Braithwaite,ScientificExplanation (Cambridge:
University "Thoughtson Teleology,"
Press,1955),ch. x, and Israel Scheffler,
BritishJournalfor the Philosophyof Science, x, 36 (February1959): 265-284.
More recentexamplesinclude FranciscoAyala, "TeleologicalExplanationsin
EvolutionaryBiology,"Philosophyof Science,xxxvii, 1 (March 1970): 1-15;
Hugh Lehman,"FunctionalExplanationsin Biology,"ibid., XXXii,1 (January
1965):1-20; RichardSorabji,"Function,"PhilosophicalQuarterly, xiv, 57 (Octo-
ber 1964): 289-302; and Larry Wright,"Functions,"PhilosophicalReview,
Lxxxii, 2 (April1973): 139-168.
74'
742 THE JOURNAL OF PHILOSOPHY
tions are just that: assumptions.They have never been systemati-

cally defended;generallytheyare not defendedat all. I thinkthere
are reasonsto suspectthatadherenceto (A) and (B) has crippledthe
mostseriousattemptsto analyzefunctionalstatementsand explana-
tion,as I will argue in sectionsI and II below. In sectionin, I will
brieflydevelop an alternativeapproach to the problem. This al-
ternativeis recommendedlargely by the fact that it emerges as
theobvious approach once we take care to understandwhyaccounts
involving(A) and (B) go wrong.
II
I begin this sectionwith a critique of Hempel and Nagel. The ob-
jections are familiarfor the most part, but it will be well to have
themfreshin our minds,fortheyformthe backdrop against which
I stagemyattackon (A) and (B).
Hempel's treatmentof functionalanalysis and explanation is a
classic example of the fusion of (A) and (B). He begins by con-
sideringthe followingsingularfunction-ascribing statement:
(1) The heartbeatin vertebrateshas the functionof circulatingthe blood

throughthe organism.
He rejectsthe suggestionthat 'function'can simplybe replaced by

'effect'on the groundsthat,althoughthe heartbeathas the effectof
producingheartsounds,thisis not its function.Presuming(B) from
the start,Hempel takesthe problemto be how the effectthe having
of which is the functionof the heartbeat(circulation)is to be dis-
tinguishedfrom other effectsof the heartbeat (e.g., heartsounds).
His answeris thatcirculation,but not heartsounds,ensuresa neces-
sary condition for the "proper workingof the organism." Thus,
Hempel proposes(2) as an analysisof (1).
(2) The heartbeatin vertebrateshas the effectof circulatingthe blood,
and this ensures the satisfactionof certain conditions (supply of
nutrimentand removal of waste) which are necessaryfor the proper
workingof the organism.
As Hempel sees the matter,themain problemwith thisanalysisis

that functionalstatementsso construedappear to have no explana-
toryforce.Since he assumes(A), the problem for Hempel is to see
whether(2) can be construedas a deductivenomologicalexplanans
for the presenceof the heartbeatin vertebrates,and, in general,to
see whetherstatementshaving the formof (2) can be construedas
deductive nomological explananda for the presencein a systemof
some traitor item thatis functionallycharacterized.
FUNCTIONAL ANALYSIS 743
Suppose,then,thatwe areinterested ofa

in explainingtheoccurrence
traiti in a system s (at a certaintimet),and thatthefollowing func-
tionalanalysisis offered:
(a) At t,s functions adequatelyin a settingofkindc (characterized
byspecific internaland externalconditions).
(b) s functions adequatelyin a settingof kindc onlyif a certain
necessary condition, n,is satisfied.
conditionn would
(c) If traiti werepresentin s then,as an effect,
be satisfied.
(d) Hence,at t, traiti is presentin s (Hempel,310).
(d), of course,does not followfrom(a)-(c), since some traiti' differ-
ent fromi mightwell suffice forthe satisfactionof conditionn. The
argumentcan be patched up by changing(c) to (c'): "Condition n
would be satisfiedin s only if traiti werepresentin s," but Hempel
rightlyrejectsthis avenue on the groundsthat instancesof the re-
sultingschemawould typicallybe false.It is false,forexample, that
theheartis a necessaryconditionforcirculationin vertebrates, since
artificialpumps can be, and are, used to maintainthe flowof blood.
We are thus leftwith a dilemma. If the original schema is correct,
thenfunctionalexplanationis invalid. If the schemais revisedso as
to ensure the validity of the explanation, the explanation will
typicallybe unsound,havinga falsethirdpremise.
Ernest Nagel offersa defenseof what is substantiallyHempel's
schemawith(c) replacedby (c').
. . . a teleological
statementof theform,"The function of A in a sys-
temS withorganization C is to enable S in the environment E to
engagein processP," can be formulated moreexplicitlyby: every
system C and in environment
S withorganization E engagesin process
P; if S withorganization C and in environment E does not have A,
thenS doesnotengagein P; hence,S withorganization C musthave
A (Nagel,403).
Thus he suggeststhat(3) is to be renderedas (4):
(3) The function in plantsis to enable themto perform
of chlorophyll
photosynthesis.
conditionfortheoccurrence
(4) A necessary ofphotosynthesisin plantsis
thepresenceofchlorophyll.
So Nagel mustface thesecondhorn of Hempel's dilemma: (3) is pre-
sumably true,but (4) may well be false. Nagel is, of course,aware
of this objection. His rathercurious responseis that,as far as we
know,chlorophyllis necessaryforphotosynthesis in the greenplants
(404). This may be so, but the responsewill not survivea change
of example. Hearts are not necessaryforcirculation,artificialpumps
having actually been incorporatedinto the circulatorysystemsof

vertebratesin such a way as to preservecirculationand life.
A more promisingdefense of Nagel might run as follows.Al-
though it is true that the presence of a workingheart is not a
necessarycondition of circulationin vertebratesunder all circum-
stances,still, under normal circumstances-mostcircumstancesin
fact-a workingheartis necessaryforcirculation.Thus it is perhaps
true that, at the presentstage of evolution,a vertebratethat has
not been tamperedwithsurgicallywould exhibitcirculationonly if
it were to contain a heart.If thesecircumstancesare specificallyin-
cluded in the explanans,perhaps we can avoid Hempel's dilemma.
Thus, insteadof (4) we should have:
a necessary
(4') At thepresentstageof evolution, conditionforcircula-
tionin vertebrates
thathavenotbeensurgicallytampered withis the
operationof a heart (properlyincorporatedinto the circulatory
system).
(4'), in conjunctionwithstatementsassertingthata givenvertebrate
exhibitscirculationand has not been surgicallytamperedwith and
is at the presentstage of evolution,will logically imply that that
vertebratehas a heart.It seems,then,that the Hempelian objection
could be overcomeif it werepossible,givena truefunction-ascribing
statementlike (1) or (3), to specify"normalcircumstances"in such a
way as to make it true that,in thosecircumstances, the presenceof
the itemin question is a necessaryconditionfor the performanceof
thefunctionascribedto it.
This defensehas some plausibilityas long as we stickto theusual
examples drawn from biology. But if we widen our view a bit,
even within biology,I think it can be shown that this defenseof
Nagel's positionwill not suffice. Considerthe kidneys.The function
of the kidneysis to eliminatewastesfromthe blood. In particular,
the functionof my leftkidneyis to eliminatewaste frommy blood.
Yet the presenceof my leftkidneyis not, in normal circumstances,
a necessaryconditionforthe removalof the relevantwastes.Only if
somethingseriouslyabnormal should befall my rightkidneywould
the operation of my left kidney become necessary,and this only
on the assumptionthat I am not hooked up to a kidneymachine."
1 It might be objected here that, although it is the functionof the kidneys to
eliminate waste, that is not the functionof a particular kidney unless operation
of that kidney is necessary for removal of wastes. But suppose scientistshad
initially been aware of the existence of the left kidney only. Then, on the
account being considered, anything they had said about the function of that
organ would have been false, since, on that account, it has no function in
havingtwokidneys!
organisms
A less obvious counterexamplederivesfromthe well-attested fact

of hemisphericalredundancyin the brain. No doubt it is in prin-
ciple possible to specifyconditionsunder which a particulardupli-
cated mechanism would be necessaryfor normal functioningof
the organism,but (a) in mostcases we are not in a positionactually
to do this,thoughwe are in a position to make well-confirmed state-
ments about the functionsof some of these mechanisms,and (b)
thesecircumstancesare by no means the normal circumstances.In-
deed, given the fact that each individual nervoussystemdevelops
somewhatdifferently owing to differingenvironmentalfactors,the
circumstances in question might well be differentfor each indi-
vidual,or forthesame individual at different times.
ApparentlyNagel was pursuingthe wrongstrategyin attempting
to analyze functionalascriptionsin termsof necessaryconditions.
Indeed, we are still faced with the dilemma noticed by Hempel: an
analysisin termsof necessaryconditionsyieldsa valid but unsound
explanatoryschema; analysisin termsof sufficient conditionsalong
the lines proposed by Hempel yieldsa schema with true premises,
but validityis sacrified.
Somethinghas gone wrong,and it is not too difficult to locate the
problem. An attempt to explain the presence of something by
appeal 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 thereinstead. In itself,this is not a serious
matter.But the accounts we have been consideringassume that
explanation is a species of deductive inference,and one cannot
deduce heartsfromcirculation.This is what underliesthe dilemma
we have been considering.At best, one can deduce circulators
from circulation. If we make this amendment,however,we are
leftwith a functionallytainted analysis; 'the functionof the heart
is to circulatethe blood' is rendered'a blood circulatoris a (neces-
sary/sufficient) condition of circulation,and the heart is a blood
circulator'.The expressionin italics is surelyas much in need of
analysis as the analyzed expression.The problem, however,runs
much deeper than the fact that the performanceof a certainfunc-
tion does not determine how that function is performed.The
problem is rather that to "explain" the presence of the heart in
vertebratesby appeal to what the heart does is to "explain" its
presenceby appeal to factorsthat are causally irrelevantto its pres-
ence. Even if it were possible,as Nagel claimed, to deduce the pres-
ence of chlorophyllfrom the occurrence of photosynthesis,this
would fail to explain the presence of chlorophyllin green plants
in just the way deducing the presence and height of a building
746 THE JOURNAL OF PHIlLOSOPHY
fromthe existenceand lengthof its shadow would fail to explain

why the building is there and has the height it does. This is not
because all explanation is causal explanation: it is not. But to ex-
plain in the presenceof a naturallyoccurringstructureor physical
process-to explain why it is there,why such a thing exists in the
place (system,context)it does-this does require specifyingfactors
thatcausallydeterminethe appearance of thatstructureor process.2
There is, of course, a sense in which the question, "Why is x
there?"is answeredby giving x's function.Consider the following
exchange. X asks Y, "Why is that thing there (pointing to the
gnomon of a sundial)?" Y answers,"Because it casts a shadow on
the dial beneath, therebyindicating the time of day." It is ex-
changes of this sort that most philosophers have had in mind
when they speak of functional explanation. But it seems to me
that, although such exchanges do representgenuine explanations,
the use of functionallanguage in this sort of explanation is quite
distinctfromits explanatoryuse in science. In section iII below, I
will sketch what I think is the central explanatoryuse of func-
tional language in science. Meanwhile, if I am right,the evident
proprietyof exchanges like that imagined between X and Y has
led to prematureacceptance of (A), and hence to concentration
on what is, from the point of view of scientificexplanation, an
irrelevantuse of functionallanguage. For it seems to me that the
question,"whyis x there?"can be answeredby specifyingx's func-
tion only if x is or is part of an artifact.Y's answer,I think,ex-
plains the presenceof the gnomonbecause it rationalizesthe action
of the agent who put it thereby supplyinghis reason forputtingit
there. In general, when we are dealing with the result of a de-
liberateaction,we may explain the resultby explaining the action,
and we may explain a deliberate action by supplying the agent's
reason forit. Thus, when we look at a sundial,we assume we know
in a general way how the gnomon came to be there: someone de-
liberatelyput it there.But we may wish to know why it was put
there. Specifyingthe gnomon's function allows us to formulate
what we suppose to be the unknown agent's reason for puttingit
there:he believed it would cast a shadow such that . . . , and so on.
2 Even in the case of a designed artifact,it is at most the designer's belief that
x will performf in s which is causally relevant to x's presence in s, not x's

actually performingf in s. The nearest I can come to describing a situation in
which x performingf in s is causally relevant to x's presence in s is this:
the designer of s notices a thing like x performingf in a systemlike s, and this
leads to belief that x will performf in s, and this in turn leads the designer to
put x in s.
When we do this,we are elaborating on what we assume is the

crucial causal factorin determiningthe gnomon'spresence,namely
a certaindeliberateaction.
If this is on the righttrack,then the viabilityof the sort of ex-
planation in question should depend on the assumption that the
thingfunctionallycharacterizedis thereas the resultof deliberate
action. If that assumptionis evidentlyfalse,specifyingthe thing's
functionwill not answerthe question. Suppose it emergesthat the
sundial is not, as such, an artifact.When the ancient building was
ruined, a large stone fragmentfell on a kind of zodiac mosaic
and embedded itselfthere.Since no sign of the room remains,Y
has mistakenlysupposed the thingwas designedas a sundial. As it
happens,the local people have been using the thingto tell time for
centuries;so Y is rightabout thefunctionof the thingX pointed to.8
But it is simply false that the thing is there because it casts a
shadow, for there is no agent who put it there "because it casts a
shadow." Again, the functionof a bowl-likedepressionin a huge
stone may be to hold holy water,but we cannot explain why it is
thereby appeal to its functionif we know it was leftthereby pre-
historicglacial activity.
If this is right,then (A) will lead us to focus on a typeof expla-
nation which will not apply to natural systems:chlorophylland
hearts are not "there" as the result of any deliberate action, and
hence the essentialpresuppositionof the explanatorymove in ques-
tion is missing.Once this becomes clear, to continue to insist that
there must be some sense in which specifyingthe function of
chlorophyllexplains its presence is an act of desperation born
of thinkingthereis no otherexplanatoryuse of functionalcharac-
terizationin science.
Why have philosophers identifiedfunctional explanation ex-
clusivelywith the appeal to something'sfunctionin explainingwhy
it is there?One reason, I suspect,is a failure to distinguishteleo-
logical explanation fromfunctionalexplanation, perhaps because
functional concepts do loom large in "explanations" having a
teleologicalform.Someone who fails to make this distinction,but
who senses that thereis an importantand legitimateuse of func-
3 Is casting a shadow the functionof this fragment?Standard use may confera
function 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 a
door stop, then the function of that block is to hold my door open. If non-
artifactsever have functions,appeals to those functions cannot explain their
presence. The things functionally characterized in science are typically not
artifacts.
tional characterizationin scientificexplanation, will see the prob-

lem as one of findinga legitimateexplanatoryrole for functional
characterizationwithin the teleological form.Once we leave arti-
factsand go to naturalsystems, however,thisapproach is doomed to
failure,as criticsof teleologyhave seen forsome time.
This mistakeprobablywould have sorteditselfout in timewere
it not thatwe do reason fromthe performanceof a functionto the
presence of certain specific processes and structures,e.g., from
photosynthesis to chlorophyll,or fromcoordinatedactivityto nerve
tissue.This is perfectly legitimatereasoning:it is a speciesof infer-
ence to the best explanation. Our best (only) explanation of photo-
synthesisrequires chlorophyll,and our best explanation of coordi-
nated activityrequires nerve tissue. But once we see what makes
this reasoninglegitimate,we see immediatlythat inferenceto an
explanation has been mistakenfor an explanation itself.Once this
becomes clear, it becomes equally clear that (A) has mattersre-
versed: given that photosynthesis is occurringin a particularplant,
we may legitimatelyinferthat chlorophyllis presentin that plant
preciselybecause chlorophyllentersinto our best (only) explanation
of photosynthesis, and givencoordinatedactivityon the part of some
animal, we may legitimatelyinferthat nerve tissue is presentpre-
cisely because nerve tissue entersinto our best explanation of co-
ordinatedactivityin animals.
To attemptto explain the heart'spresencein vertebratesby ap-
pealing to its functionin vertebratesis to attemptto explain the
occurrenceof heartsin vertebratesby appealing to factorsthat are
causallyirrelevantto its presencein vertebrates. This facthas given
"functionalexplanation" a bad name. But it is (A) thatdeservesthe
blame. Once we see (A) as an undefendedphilosophical hypothe-
sis about how to construefunctionalexplanationsratherthan as a
statementof the philosophical problem, the correctalternativeis
obvious: what we can and do explain by appeal to what something
does is thebehaviorof a containingsystem.4
A much more promisingsuggestionin the lightof theseconsider-
ations is that (1) is appealed to in explaining circulation.If we
reject (A) and adopt thissuggestion,a simple deductive-nomological
explanation with circulationas the explicandum turnsout to be a
sound argument.
4 A confused perception of this fact no doubt underlies (B), but the fact that
(B) is nearly inseparable from (A) in the literature shows how confused this
perceptionis.
(5) a. Vertebratesincorporating a beatingheartin theusualway(in the

ways does)exhibitcirculation.
b. Vertebrates incorporatesa beatingheartin theusualway.
c. Hence,s exhibitscirculation.
Though by no means flawless,(5) has severalvirtues,not the least

of which is that it does not have biologistspassing by an obvious
application of evoluation or genetics,in favorof an invalid or un-
sound "functional" explanation of the presence of hearts. Also,
the redundancyexamples are easily handled; e.g., the removal of
wastesis deduced in thekidneycase.
The implausibilityof (A) is obscured in examples taken from
biologyby the factthat thereare two distinctuses of functionstate-
mentsin biology.Considerthefollowingstatements.
(a) The function vacuolein protozoans
of thecontractile is elimination
ofexcesswaterfromtheorganism.
(b) The functionof theneurofibrils of the
in theciliatesis coordination
ofthecilia.
activity
These statementscan be understoodin eitherof two ways.(i) They

are generally used in explaining how the organism in question
comes to exhibit certain characteristicsor behavior. Thus (a) ex-
plains how excesswater,accumulatedin the organismby osmosis,is
eliminatedfromthe organism;(b) explains how it happens that the
activityof the cilia in paramecium,forinstance,is coordinated.(ii)
They may be used in explaining the continued survivalof certain
organismsincorporatingstructuresof the sort in question by in-
dicatingthe survivalvalue that would accrue to such organismsin
virtueof having structuresof that sort.Thus (a) allows us to infer
that incorporationof a contractilevacuole makes it possible for
the organism to be surrounded by a semi-permeablemembrane,
allowing the passage of oxygeninto, and the passage of wastesout
of, the organism.Relatively free osmosis of this sort is obviously
advantageous,and this is made possible by a structurewhich solves
the excess water problem. Similarly,ciliates incorporatingneuro-
fibrilswill be capable of fairlyefficientlocomotion, the survival
value of whichis obvious.'
5 Notice that the second use is parasitic on the first.It is only because the
neurofibrilsexplain the coordinated activityof the cilia that we can assign a
survival value to neurofibrils:the survival value of a structures hangs on what
capacities of the organism, if any, are explicable by appeal to the function-
ing of s.
The second sortof use occurs as part of an account which,if we

are not careful,can easily be mistakenfor an explanation of the
presence of the sort of item functionallycharacterized,and this
has perhaps encouraged philosophersto accept (A). For it might
seem thatnatural selectionprovidesthe missingcausal link between
what somethingdoes in a certain type of organismand its pres-
ence in that typeof organism.By performingtheirrespectivefunc-
tions, the contractilevacuole and the neurofibrilshelp species in-
corporatingthem to survive,and therebycontributeto their own
continued presence in organismsof those species, and this might
seem to explain the presenceof those structuresin the organisms
incorporatingthem.
Plausible as this sounds, it involves a subtle yet fundamental
misunderstandingof evolutionarytheory.A clue to the mistakeis
found in the fact that the contractilevacuole occurs in marine
protozoansthat have no excess-water problembut the reverseprob-
lem. Thus the functionand effecton survival of this structureis
not thesame in all protozoans.Yet the explanationof its presencein
marine and fresh-water species is almost certainlythe same. This
fact remindsus that the processesactually responsiblefor the oc-
currenceof contractilevacuoles in protozoans are totally insensi-
tive to what that structuredoes. Failure to appreciate this point
not only lends spurious plausibilityto (A) as applied to biological
examples; it seriouslydistortsour understandingof evolutionary
theory.Whetheran organismo incorporatess depends on whether
s is "specified"by the genetic "plan" which o inheritsand which,
of o's species.Altera-
at a certainlevel of abstraction,is characteristic
tionsin the plan are due to mutation.If a plan is alteredso that it
specifiess' rather than s, then the organismsinheritingthis plan
will incorporatest, regardlessof the functionor survival value of
s' in thoseorganisms.If the alterationis advantageous,the number
of organismsinheritingthat plan may increase,and, if it is disad-
vantageous,theirnumbermay decrease.But thishas no effectwhat-
everon the plan, and therefore no effectwhateveron the occurrence
ofstin theorganismsin question.
One sometimeshears it said that natural selectionis an instance
of negativefeedback.If thisis meant to implythat the relativesuc-
cess or failure of organismsof a certain type can affecttheir in-
heritedcharacteristics, it is simplya mistake: the characteristics of
organismswhich determinetheir relative success are determined
by their genetic plan, and the characteristicsof these plans are
utterlyindependent of the relative success of organismshaving
FUNCrIONAL ANALYSIS 751
them. Of course, if s is verydisadvantageousto organismshaving

a plan specifyings, then organismshaving such plans may dis-
appear altogether,and s will no longeroccur. We could, therefore,
thinkof natural selectionas reactingon the set of plans generated
by mutationby weedingout the bad plans: natural selectioncannot
alter a plan, but it can trimthe set. Thus, we may be able to ex-
plain why a given plan is not a failure by appeal to the functions
of the structuresit specifies.Perhaps this is what some writershave
had in mind.But thisis not to explain why,e.g.,contractilevacuoles
occur in certainprotozoans,it is to explain why the sort of proto-
zoan incorporatingcontractilevacuoles occurs.Since we cannot ap-
peal to the relativesuccessor failureof theseorganismsto explain
whytheirgeneticplan specifiescontractilevacuoles, we cannot ap-
peal to the relativesuccessor failureof theseorganismsto explain
whytheyincorporatecontractilevacuoles.
Once we are clear about the explanatoryrole of functionsin
evolutionarytheory,it emerges that the function of an organ
or process (or whatever)is appealed to to explain the biological
capacities of the organismcontainingit, and fromthese capacities
conclusionsare drawn concerningthe chances of survivalfor orga-
nismsof that type.For instance,appeal to the functionof the con-
tractilevacuole in certainprotozoansexplains how theseorganisms
are able to keep fromexploding in freshwater.Thus evolutionary
biologydoes not providesupportfor (A) but for the idea instanced
in (5): identifyingthe functionof somethinghelps to explain the
capacitiesof a containingsystem.6
(A) misconstrues functionalexplanationby misidentifying what is
explained. Let us abandon (A), then, in favor of the view that
functionsare appealed to in explaining the capacities of contain-
ing systems,and turnour attentionto (B).
Whereas (A) is a thesis about functionalexplanation, (B) is a
thesisabout the analysisof function-ascribing statements.Perhaps
when divorced from(A), as it is in (5), it will fare better than it
does in the accountsof Hempel and Nagel.
6 In addition to the misunderstandingabout evolutionary theory just dis-

cussed, biological examples have probably suggested (A) because biology was the
locus classicus of teleological explanation. This has perhaps encouraged a con-
fusionbetween the teleologicalformof explanation, incorporatedin (A), with the
explanatory role of functional ascriptions. Function-ascribing statements do
occur in explanations having a teleological form, and, when they do, their
interest is vitiated by the incoherence of that form of explanation. It is the
legitimateuse of function-ascribing statementsthat needs examination, i.e., their
contributionto nonteleological theoriessuch as the theoryof evolution.
I1
In spite of the evidentvirtuesof (5), (5a) has seriousshortcomings as
an analysisof (1). In factit is subject to the same objection Hempel
brings to the analysis that simply replaces 'function' by 'effect':
vertebratesincorporatinga workingheart in the usual way exhibit
the productionof heartsounds,yetthe productionof heartsoundsis
not a function of hearts in vertebrates.The problem is that,
whereas the productionof certain effectsis essential to the heart's
performingits function, there are some effectsproduction of
which is irrelevantto the functioningof the heart. This problem
is bound to infect any "selected-effects" theory,i.e., any theory
built on (B).
What is needed to establisha selected-effects theoryis a general
formulathat identifies the appropriate Both Hempel and
effects.7
Nagel attemptto solve this problem by identifying the functionof
somethingwith just those effects which contribute to the mainte-
nance of some special conditionof, or the performance of some spe-
cial activityof, some containingsystem. this If sort of solution is
to be viable, there must be some principled way of selectingthe
relevantactivitiesor conditionsof containingsystems.For no mat-
ter which effectsof somethingyou happen to name, therewill be
some activityof thecontainingsystemto whichjust thoseeffects con-
tribute,or some conditionof the containingsystemwhich is main-
tained with the help of just those effects.Heart activity,for ex-
ample, keeps the circulatorysystemfrombeing entirelyquiet, and
the appendix keeps people vulnerable to appendicitis.8
Hempel suggeststhat,in general,the crucial featureof a contain-
ing system,contributionto whichis to count as the functioningof a
7 Larry Wright (op. cit.) is aware of this problem, but does not, to my mind,
make much progress with it. Wright's analysis rules out "The function of the
heart is to produce heart sounds," on the ground that the heart is not there
because it produces heartsounds. I agree. But neither is it there because it
pumps blood. Or if, as Wright maintains, there is a sense of 'because' in which
the 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'. Wright does not attempt to provide such an analysis, but depends on the
fact that, in many cases, we are able to use the word in the required way.
But we are also able to use 'function' correctlyin a varietyof cases. Indeed, if
Wright is right,the words are simply interchangeablewith a little grammatical
maneuvering. The problem is to make the conditions of correct use explicit.
Failure to do this means that Wright'sanalysis providesno insightinto the prob-
lem of how functional theories are confirmed,or whence they derive their
explanatoryforce.
8 Surprisingly,when Nagel comes to formulate his general schema of func-
tional attributionhe simply ignores this problem, and thus leaves himself open
to the trivializationjust suggested. Cf., Nagel, op. cit., p. 403.
FUNCrIONAL ANALYSIS 753
contained part, is that the systembe maintained in "adequate, or

effective,or proper workingorder" (306). Hempel explicitly de-
clines to discuss what constitutes"proper working order," pre-
sumablybecause he rightlythinksthat thereare more seriousprob-
lems with the analysis he is discussingthan those introduced by
this phrase. But it seemsclear that forsomethingto be in working
orderis just forit to be capable of performing its functions,and for
it to be in adequate or effectiveor properworkingorder is just for
it to be capable of performing its functionsadequately or effectively
or properly.Hempel seemsto realize thishimself,forin settingforth
a deductiveschemaforfunctionalexplanation,he glossesthe phrase
in question as 'functionsadequately' (310). More generally,if we
identifythe functionof somethingx with those effectsof x which
contributeto the performanceof some activitya or to the mainte-
nance of some condition c of a containingsystems, then we must
be preparedto say as well that a functionof s is to performa or to
maintain c. This suggeststhe followingformulationof "selected-
effects" theories.
(6) The function ofan F in a G is fjustin case (thecapacityfor)f is an
effect
of an F incorporated in a G in theusual way(or: in theway
thisF is incorporatedin thisG), and thateffect contributesto the
performance ofa function of thecontaining G.
It seems that any theorybased on (B)-what I have been calling
theories-must ultimatelyamount to something
"selected-effects"
like (6).9 Yet (6) cannot be the whole story about functional
ascriptions.
Suppose we follow(6) in rendering,"The functionof the contrac-
tile vacuole in protozoansis eliminationof excess water from the
organism."The resultis (7):
(7) Eliminationof excesswaterfromtheorganism of a con-
is an effect
tractile intheusualwayin a protozoan,
vacuoleincorporated and that
effect
contributes of a function
to the performance of a protozoan.
In orderto test(7) we should have to know a statementof the form,
"f is a functionof a protozoan."Perhaps protozoanshave no func-
tions. If not, (7) is just a mistake.If theydo, then presumablywe
shall have to appeal to (6) foran analysisof the statementattribut-
ing such a function,and thiswill leave us with anotherunanalyzed
functionalascription.Either we are launched on a regress,or the
9 Hugh Lehman (op. cit.) given an analysisthat appears to be essentially
like(6).
analysisbreaksdown at some level forlack of functions,or perhaps

forlack of a plausible condidateforcontainingsystems.If we do not
wish simplyto acquiesce in the autonomyof functionalascriptions,
it must be possible to analyze at least some functionalascriptions
withoutappealing to functionsof containingsystems.If (6) can be
shown to be the only plausible formulationof theoriesbased on
(B), then no such theorycan be the whole story.
Our question,then,is whethera thing'sfunctioncan plausiblybe
identifiedwith those of its effectscontributingto production of
some activityof,or maintenanceof some conditionof, a containing
system,where performanceof the activityin question is not a
function of the containing system.Let us begin by considering
Hempel's suggestionthat functionsare to be identifiedwith pro-
duction of effectscontributingto proper working order of a
containingsystem.I claimed earlier on that to say somethingis in
proper workingorder is just to say that it properlyperformsits
functions.This is fairlyobvious in cases of artifactsor tools. To
make a decision about which sort of behavior counts as working
amounts to deciding about the thing'sfunction.To say something
is working,though not behaving or disposed to behave in a way
having anythingto do with its function,is to be open, at the
veryleast,to thechangeof arbitrariness.
When we are dealing with a living organism, or a society
of living organisms,the situationis less clear. If we say,"The func-
tion of the contractilevacuole in protozoansis eliminationof excess
water from the organism,"we do make referenceto a contain-
ing organism,but not, apparently,to its function(if any). How-
ever,since contractilevacuoles do a numberof thingshavingnothing
to do with theirfunction,theremust be some implicitprincipleof
selection at work. Hempel's suggestionis that, in this context,to
be in "properworkingorder"is simplyto be alive and healthy.This
worksreasonablywell for certainstandardexamples, e.g., (1) and
(3): circulation does contributeto health and survival in verte-
brates,and photosynthesis does contributeto health and survivalin
greenplants.10But, once again, theprinciplewill not standa change
of example, even within the life sciences.First,there are cases in
which proper functioningis actually inimical to health and life:
10 Even these applications have their problems. Frankfurtand Poole ["Func-

tional Explanations in Biology," British Journal for the Philosophy of Science,
xvii (1966)] point out that heartsounds contribute to health and survival via
theirusefulnessin diagnosis.
FUNCTIONALI ANALYSIS 755
functioningof the sex organsresultsin the death of individualsof

many species (e.g., certainsalmon). Second, a certainprocessin an
organism may have effectswhich contribute to health and sur-
vival but which are not to be confusedwith the functionof that
process:secretionof adrenalin speeds metabolismand therebycon-
tributesto eliminationof harmfulfat deposits in overweighthu-
mans,but thisis not a functionof adrenalinsecretionin overweight
humans.
A more plausible suggestionalong these lines in the special con-
textof evolutionarybiologyis this:
(8) The functionsofa partor processin an organism are to be identified
withthoseof itseffects
contributingto activities
or conditions of the
organismwhichsustainor increasethe organism's capacityto con-
tribute
to survival
ofthespecies.
Give or take a nicety,(8) doubtlessdoes capture a great many uses
of functionallanguage in biology. For instance,it correctlypicks
out eliminationof excess water as the functionof the contractile
vacuole in fresh-water protozoansonly, and correctlyidentifiesthe
functionof sex organs in species in which the exercise of these
organsresultsin the death of the individual."-
In spite of these virtues,however, (8) is seriously misleading
and extremelylimited in applicability even within biology. Evi-
dently,what contributesto an organism'scapacity to maintain its
species in one sort of environmentmay undermine that capacity
in another. When this happens, we might say that the organ (or
whatever)has lost its function.This is probablywhat we would say
about the contractilevacuole if fresh-waterprotozoanswere success-
fully introducedinto salt water,for in this case the capacity ex-
plained would no longerbe exercised.But if the capacityexplained
by appeal to the functionof a certain structurecontinued to be
exercisedin the new environment,thoughnow to the individual's
detriment,we would not say that that structurehad lost its func-
tion. If, forsome reason,flyingceased to contributeto the capacity
of pigeons to maintain their species, or even undermined that
capacityto some extent,'2we would still say that a functionof the
wingsin pigeonsis to enable themto fly.Only if the wingsceased to
11 Michael Ruse has argued for a formulation like (8). See "Function State-
ments in Biology," Philosophy of Science, xxxviii, 1 (March 1971): 87-95, and
The Philosophy of Biology (London: Hutchinson, 1973).
12 Perhaps, in the absence of serious predators,with a readily available food
supply, and with no need to migrate,flyingsimply wastes energy.
756 THkE JOURNAL OF PHILOSOPHY
functionas wings,as in the penguins or ostriches,would we cease

to functionallyanalyze skeletal structureand the like with an eye
to explaining flight.Flight is a capacitythat cries out for explana-
tion in termsof anatomical functionsregardlessof its contribution
to thecapacityto maintainthespecies.
What thisexample showsis thatfunctionalanalysiscan properly
be carriedon in biology quite independentlyof evolutionarycon-
siderations:a complex capacityof an organism(or one of its parts
or systems)may be explained by appeal to a functionalanalysisre-
gardlessof how it relatesto the organism'scapacityto maintain the
species.At best,then,(8) picks out thoseeffects which will be called
functionswhen what is in the offingis an application of evolution-
ary theory.As we shall see in the next section,(8) is misleadingas
well in that it is not which effectsare explained but the styleof
explanation that makes it appropriate to speak of functions.(8)
simplyidentifieseffects which,as it happens, are typicallyexplained
in thatstyle.
We have not quite exhausted the lessonsto be learned from(8).
The plausibilityof (8) restson theplausibilityof the claim that,for
certainpurposes,we may assume that a functionof an organismis
to contributeto thesurvivalof its species.What (8) does, in effect, is
identifya functionof an importantclass of (uncontained)containing
systemswithoutprovidingan analysisof the claim thata functionof
an organismis to contributeto thesurvivalof itsspecies.
Of course, an advocate of (8) might insist that it is no part of
his theoryto claim thatmaintenanceof the species is a functionof
an organism.But then the defenseof (8) would have to be simply
that it describesactual usage, i.e., that it is in facteffectscontribut-
ing to an organism'scapacity to maintain its species which evolu-
tionarybiologistssingleout as functions.Construedin thisway, (8)
would, at most,tell us which effectsare picked out as functions;it
would provide no hint as to why these effectsare picked out as
functions.We know why evolutionarybiologistsare interestedin
effectscontributingto an organism'scapacityto maintainits species,
but why call them functions?This is preciselythe sort of ques-
tion that a philosophical account of function-ascribing statements
should answer.Either (8) is defendedas an instanceof (6)-mainte-
nance of the species is declared a functionof organisms-or it is
defendedas descriptiveof usage. In neithercase is any philosophical
analysisprovided. For in the firstcase (8) relies on an unanalyzed
(and undefended)function-ascribing statement,and in the second it
failsto giveany hintas to the pointof identifying certaineffects

as functions.
The failingsof (8) are,I think,boundto crippleanytheorythat
identifiesa thing'sfunctionswitheffectscontributing to someante-
cedentlyspecifiedtypeof conditionor behaviorof a containing
system. If the theoryis an instanceof (6), it launchesa regressor
terminates in an unanalyzedfunctionalascription;if it is not an
instanceof (6), thenit is boundto leaveopen theveryquestionat
issue,viz.,whyare theselectedeffectsseenas functions?
III
In thissection,
I willsketchbriefly an accountoffunctional explana-
tionwhichtakesseriously the intuitionthatit is a genuinelydis-
tinctivestyleofexplanation. The assumptions (A) and (B) formthe
core of approachesthatseek to minimizethe differences between
functional explanationsand explanationsnot formulated in func-
tionalterms. Suchapproacheshavenotgivenmuchattention to the
characterizationof thespecialexplanatory strategy scienceemploys
in usingfunctional language,fortheproblemas it was conceived
in suchapproacheswas to showthatfunctional explanationis not
reallydifferentin essentialsfromotherkindsof scientific explana-
tion. Once the problemis concenvedin thisway,one is almost
certainto missthe distinctive featuresof functional explanation,
and henceto missthepointoffunctional description. The account
ofthissectionreverses thistendency byplacingprimary emphasis on
thekindofproblemthatis solvedbyappeal to functions.
1. Functionsand Dispositions.Something marbe capableofpump-

ing even thoughit does not functionas a pump (ever)and even
thoughpumpingis notitsfunction. On theotherhand,ifsomething
functions as a pumpin a system s or if thefunctionof something
in a systems is to pump,thenit mustbe capableofpumpingin s."s
13 ThroughoutthissectionI am discounting appeals to the intentions of de-
signersor users.x may be intendedto preventaccidentswithoutactuallybeing
capable of doing so. With referenceto this intentionit would be properin
certaincontextsto say,"x's functionis to preventaccidents,thoughit is not
actuallycapableofdoingso."
There can be no doubt thata thing'sfunctionis oftenidentified withwhat
it is typically
or "standardly"used to do, or withwhatit was designedto do.
But the sortsof thingsforwhichit is an important scientific problemto pro-
vide functionalanalyses-brains,organisms, societies,social institutions-either
do not have designersor standardor regularuses at all, or it would be in-
appropriateto appeal to thesein constructing and defininga scientific theory
becausethe designeror use is not known-brains,devicesdug up by archaeolo-
Thus, function-ascribing statementsimply disposition statements;

to attributea function to somethingis, in part, to attribute a
dispositionto it. If the functionof x in s to p, then x has a disposi-
tion to d in s. For instance,if the functionof the contractilevacuole
in fresh-water protozoans is to eliminate excess water from the
organism,then theremust be circumstancesunder which the con-
tractilevacuole would actually manifesta dispositionto eliminate
excesswaterfromtheprotozoanthatincorporatesit.
To attributea dispositiond to an object a is to assert that the
behavior of a is subject to (exhibits or would exhibit) a certain
lawlike regularity:to say a has d is to say that a would manifest
d (shatter,dissolve) were any of a certainrange of events to occur
(a is put in water,a is strucksharply).The regularityassociated
with a disposition-call it the dispositionalregularity-is a regu-
laritythat is special to the behaviorof a certainkind of object and
obtains in virtueof some special fact(s)about that kind of object.
Not everything is water-soluble:such thingsbehave in a special way
in virtue of certain (structural)featuresspecial to water-soluble
things.Thus it is that dispositionsrequire explanation: if x has d,
thenx is subject to a regularityin behaviorspecial to thingshaving
d, and such a factneeds to be explained.
To explain a dispositional regularityis to explain how mani-
festationsof the dispositionare broughtabout given the requisite
precipitatingconditions.In what follows,I will describetwodistinct
strategiesforaccomplishingthis.It is mycontentionthat the appro-
priatenessof function-ascribing statementscorrespondsto the ap-
propriatenessof the -secondof these two strategies.This, I think,
explains the intuitionthat functionalexplanation is a special kind
of explanation.
2. Two ExplanatoryStrategies
(i) The SubsumptionStrategy.Suppose a has a dispositiond. The
associated dispositional regularityconsistsin the fact that certain
kinds of eventswould cause a to manifestd. One way to explain
thisfactwould be to discoversome featureof a which allowed us to
representthe connection between precipitatingevents and mani-
gists-or because there is some likelihood that real and intended function
diverge-social institutions,complex computers.Functional talk may have origi-
nated in contexts in which referenceto intentions and purposes loomed large,
but referenceto intentions and purposes does not figure at all in the sort of
functionalanalysis favoredby contemporarynatural scientists.
festationsas instancesof one or moregeneral laws, i.e., laws govern-

ing the behaviorof thingsgenerally,not just thingshaving d. Brian
O'Shaughnessyhas providedan example whichallows a particularly
simple illustrationof this strategy.'4Consider the disposition he
calls elevancy: the tendencyof an object to rise in water of its own
accord. To explain elevancy,we must explain why freeinga sub-
mergedelevantobject causes it to rise.15This we may do as follows.
In every case, the ratio of an elevant object's mass to its non-
permeablevolume is less than the density(mass per unit volume) of
water.Archimedes'principle tells us that water exertsan upward
forceon a submergedobject equal to the weightof the water dis-
placed. In the case of an elevantobject, thisforceevidentlyexceeds
the weight of the object by some amount f. Freeing the object
changes the net forceon it fromzero to a net forceof magnitude
f in the directionof the surface,and the object rises accordingly.
Here, we subsume the connectionbetween freeingsand risingsun-
der a general law connectingchanges in net forcewith changes in
motion by citing a featureof elevant objects which allows us (via
Archimedes'principle)to representfreeingthemunder wateras an
instanceof introducinga net forcein the directionof the surface.
(ii) The Analytical Strategy.Rather than subsume a disposi-
tional regularityunder a law not special to the disposed objects,
the analytical strategyproceeds by analyzinga disposition d of a
into a numberof other dispositionsd1 . . . dn,had by a or compo-
nentsof a such that programmedmanifestationof the di resultsin
or amountsto a manifestationof d.16 The two strategieswill fitto-
getherinto a unifiedaccount if the analyzingdispositions(the di)
can be made to yieldto thesubsumptionstrategy.
When the analyticalstrategyis in the offing, one is apt to speak
of capacities (or abilities) rather than of dispositions.This shift
in terminologywill put a more familiar face on the analytical
14 "The Powerlessnessof Dispositions," Analysis, xxxi.1, 139 (October 1970):
1-15. See also my discussionof this example in "Dispositions, States and Causes,"
ibid., xxxiv.6, 162 (June 1974): 194-204.
15 Also, we must explain why submerging a free elevant object causes it to
rise, and why a free submerged object's becoming elevant causes it to rise. One
of the convenient featuresof elevancy is that the same considerationsdispose of
all these cases. This does not hold generally: gentle rubbing, a sharp blow, or a
sudden change in temperaturemay each cause a glass to manifesta disposition
to shatter,but the explanations in these cases are significantlydifferent.
16 By "programmed" I simplymean organized in a way that could be specified
in a program or flow chart: each instruction (box) specifiesmanifestationof

one of the d4 such that, if the program is executed (the chart followed), a
manifestsd.
strategy,17 for we often explain capacities by analyzing them. As-

sembly-lineproductionprovides a transparentexample of what I
mean. Productionis broken down into a number of distincttasks.
Each point on the line is responsiblefora certaintask,and it is the
functionof the workers/machines at that point to complete that
task.If the line has the capacityto produce the product,it has it in
virtueof the fact that the workers/machines have the capacities to
performtheirdesignatedtasks,and in virtueof the fact that when
thesetasksare performedin a certainorganizedway-according to a
certainprogram-the finishedproductresults.Here we can explain
the line's capacity to produce the product-i.e., explain how it is
able to produce the product-by appeal to certaincapacitiesof the
workers/machinesand their organization into an assembly line.
Againstthis background,we may pick out a certaincapacityof an
individual exerciseof whichis his functionon the line. Of the many
thingshe does and can do, his functionon the line is doing what-
ever it is thatwe appeal to in explaining the capacityof the line as
a whole. If the line producesseveralproducts,i.e., if it has several
capacities,then,althougha certaincapacityc of a workeris irrele-
vant to one capacityof the line, exerciseof c by that workermay
be his functionwith respect to another capacity of the line as a
whole.
Schematicdiagramsin electronicsprovide anotherobvious illus-
tration.Since each symbolrepresentsany physicalobject whatever
having a certaincapacity,a schematicdiagramof a complex device
constitutesan analysisof the electroniccapacitiesof the device as a
whole into the capacitiesof its components.Such an analysisallows
us to explain how the device as a whole exercisesthe analyzed ca-
pacity,for it allows us to see exercisesof the analyzed capacity as
programmedexercise of the analyzingcapacities. In this case, the
"'program"is given by the lines indicatinghow the componentsare
hooked up. (Of course,the lines are themselvesfunctionsymbols.)
Functional analysis in biology is essentiallysimilar. The bio-
logically significantcapacities of an entireorganismare explained
17Some mightwant to distinguish betweendispositionsand capacities,and
argue thatto ascribea functionto x is in part to ascribea capacity to x, not
a dispositionas I haveclaimed.Certainly (1) is strainedin a way(2) is not.
(1) Heartsare disposedto pump.
Heartshavea disposition to pump.
Sugaris capableof dissolving.
Sugarhas a capacityto dissolve.
(2) Heartsare capableofpumping.
Heartshavea capacityto pump.
Sugaris disposedto dissolve.
Sugarhas a dispositionto dissolve.
FUNCTIONAL ANALYSIS 76I
by analyzing the organisminto a number of "systems"-the circu-

latorysystem,thedigestivesystem,the nervoussystem,etc.,-each of
whichhas its characteristic capacities.1SThese capacitiesare in turn
analyzedinto capacitiesof componentorgansand structures. Ideally,
this strategyis pressed until pure physiologytakes over, i.e., until
the analyzingcapacities are amenable to the subsumptionstrategy.
We can easilyimaginebiologistsexpressingtheiranalysesin a form
analogous to the schematicdiagramsof electricalengineering,with
special symbolsforpumps,pipes, filters, and so on. Indeed, analyses
of even simple cognitivecapacities are typicallyexpressedin flow
chartsor programs,formsdesignedspecificallyto representanalyses
of informationprocessingcapacitiesgenerally.
Perhaps themostextensiveuse of the analyticalstrategyin science
occursin psychology,for a large part of the psychologist's job is to
explain how the complex behavioral capacities of organismsare
acquired and how theyare exercised.Both goals are greatlyfacili-
tatedby analysisof the capacitiesin question,forthenacquisitionof
the analyzed capacity resolvesitselfinto acquisition of the analyz-
ing capacities and the requisite organization,and the problem of
performanceresolvesitselfinto the problem of how the analyzing
capacities are exercised.This sort of strategyhas dominated psy-
chologyover since Watson attemptedto explain such complex ca-
pacities as the ability to run a maze by analyzingthe performance
into a series of conditioned responses,the stimulus for each re-
sponsebeing the previousresponseor somethingencounteredas the
resultof thepreviousresponse.19 Acquisitionof thecomplexcapacity
is resolvedinto a numberof distinctcases of simple conditioning,
i.e., the ability to learn the maze is resolvedinto the capacity for
stimulussubstitution,and the capacityto run the maze is resolved
into abilities to respond in certain simple ways to certain simple
stimuli. Watson's analysis proved to be of limited value, but the
analytic strategyremains the dominant mode of explanation in
behavioralpsychology.20
18 Indeed,what makessomething part of,e.g., the nervoussystemis thatits

capacitiesfigurein an analysisof the capacityto respondto externalstimuli,
coordinatemovement, etc.Thus, thereis no questionthattheglial cellsare part
of the brain,but thereis some questionas to whethertheyare part of the
nervoussystem or merelyauxiliaryto it.
19JohnB. Watson,Behaviorism (New York:Norton,1930)chs. ix and xi.
20 Writerson the philosophyof psychology, especiallyJerryFodor, have
graspedthe connectionbetweenfunctionalcharacterization and the analytical
strategyin psychologicaltheorizing, but have not applied the lesson to the
problemof functionalexplanationgenerally.The cleareststatementoccursin
J.A. Fodor,"The Appeal to Tacit Knowledgein Psychological Explanation,"this
jouRNAL, Lxv, 24 (December19,1968):627-640.
3. Functions and Functional Analysis.In the contextof an appli-

cation of the analyticalstrategy,exerciseof an analyzingcapacity
emergesas a function:it will be appropriateto say that x functions
as a p in s, or thatthefunctionof x in s is (-ing, when we are speak-
ing against the backgroundof an analytical explanation of some
capacityof s which appeals to the factthat x has a capacityto 4 in
s. It is appropriateto say thatthe heartfunctionsas a pump against
the backgroundof an analysisof the circulatorysystem'scapacityto
transportfood,oxygen,wastes,and so on, which appeals to the fact
that the heart is capable of pumping. Since this is the usual back-
ground,it goes withoutsaying,and this accounts for the fact that
"The heartfunctionsas a pump" soundsright,and "The heartfunc-
tions as a noise-maker"sounds wrong,in some context-free sense.
This effectis strengthened by the absence of any actual application
of the analyticalstrategywhichmakes use of the factthat the heart
makesnoise.2'
We can capturethisimplicitdependenceon an analyticalcontext
by enteringan explicit relativizationin our regimentedreconstruc-
tionof function-ascribing statements:
(9) x functions ofx in s is to 0) relativeto an
as a q in s (or: thefunction
analyticalaccountA of s's capacityto 'pjust in case x is capableof
0-ingin s and A appropriately and adequatelyaccountsfors's ca-
pacityto 'p by,in part,appealingto thecapacityof x to p in s.
Sometimeswe explain a capacity of s by analyzing it into other
capacities of s, as when we explain how someone ignorant of
cookeryis able to bake cakes by pointing out that he followed a
recipeeach instructionof whichrequiresno special capacitiesforits
execution.Here, we don't speak of,e.g.,stirringas a functionof the
cook, but ratherof the functionof stirring.Since stirringhas dif-
ferentfunctionsin different recipes and at different points in the
same recipe,a statementlike 'The functionof stirringthe mixture
is to keep it fromstickingto the bottom of the pot' is implicitly
relativizedto a certain (perhaps somewhatvague) recipe. To take
account of this sort of case, we need a slightlydifferentschema:
where e is an activityor behavior of a systems (as a whole), the
functionof e in s is to (p relativeto an analytical account A of s's
capacityto , just in case A appropriatelyand adequately accounts
fors's capacityto q by, in part,appealing to s's capacityto engage
in e.
21 It is sometimes function.
do have a psychological
thatheartsounds
suggested
dispositionappealingto the
In the contextof an analysisof a psychological
heart'snoise-making (e.g.,as
capacity,"The heartfunctionsas a noise-maker"
a producerof regularthumps),would not evensound odd.
(9) explains theintuitionbehind theregress-ridden (6): functional

ascriptionsdo require relativizationto a "functionalfact" about a
containingsystem,i.e., to the fact that a certaincapacityof a con-
taining systemis approximatelyexplained by appeal to a certain
functionalanalysis.And, like (6), (9) makes no provisionforspeak-
ing of the functionof an organismexcept against a background
analysisof a containingsystem(the hive, the corporation,the eco-
system).Once we see that functionsare appealed to in explain-
ing the capacities of containingsystems,and indeed that it is the
applicability of a certain strategyfor explaining these capacities
that makes talk of functionsappropriate,we see immediatelywhy
we do not speak of the functionsof uncontainedcontainers.What
(6) fails to capture is the fact that uncontained containerscan be
functionallyanalyzed,and the way in which function-analytical ex-
planation mediates the connectionbetween functionalascriptions
(x functionsas a 0, the functionof x is to q) and the capacities
of thecontainers.
4. Function-analyticalExplanation. If the account I have been

sketchingis to draw any distinctions,the availability and appro-
priatenessof analyticalexplanationsmust be a nontrivialmatter.22
So let us examine an obviouslytrivialapplication of the analytical
strategywith an eye to determiningwhetherit can be dismissedon
principledgrounds.
(10)Each partof themammaliancirculatory systemmakesits own dis-
tinctivesound,and makesit continuously. These combineto form
the "circulatory of all mammals.The mam-
noise" characteristic
maliancirculatorysystem is capableof producingthissoundat vari-
ous volumesand varioustempos.The heartbeatis responsible for
the throbbing characterof thesound,and it is thecapacityof the
heartto beatat variousratesthatexplainsthecapacityof thecircu-
latorysystem to producea variouslytempoedsound.
Everythingin (10) is, presumably,true.The question is whetherit
allows us to say that the functionof the heart is to produce a vari-
ouslytempoedthrobbingsound.28To answerthisquestion we must,
I think,get clear about the motivationfor applying the analytical
22 Of course, it might be thatthereare none but arbitrarydistinctionsto be

drawn. Perhaps (9) describes usage, and usage is arbitrary,but I am unable to
take this possibilityseriously.
23The issue is not whether (10) forces us, via (9), to say something false.
Relative to some analytical explanation, it may be true that the functionof the
heart is to produce a variously tempoed throbbing.But the availability of (10)
should not support such a claim.
strategy.For my contentionwill be that the analytical strategyis

mostsignificantly applied in cases veryunlike thatenvisagedin (10).
The explanatory interestof an analytical account is roughly
proportionalto (i) the extentto which the analyzingcapacities are
less.sophisticatedthan the analyzed capacities, (ii) the extent to
which the analyzing capacities are differentin type from the
analyzed capacities,and (iii) the relativesophisticationof the pro-
gramappealed to, i.e., the relativecomplexityof the organizationof
componentparts/processes that is attributedto the system.(iii) is
correlativewith (i) and (ii): the greaterthe gap in sophistication
and type between analyzing capacities and analyzed capacities,
the more sophisticatedthe programmust be to close the gap.
It is preciselythe width of thesegaps which,for instance,makes
automata theoryso interestingin its application to psychology.
Automata theorysupplies us with extremelypowerfultechniques
for constructingdiverse analyses of very sophisticatedtasks into
veryunsophisticatedtasks.This allows us to see how, in principle,
a mechanismsuch as the brain, consistingof physiologicallyun-
sophisticatedcomponents (relativelyspeaking), can acquire very
sophisticatedcapacities. It is the prospect of promoting the ca-
pacity to store ones and zeros into the capacity to solve logic
problemsand recognizepatternsthat makes the analyticalstrategy
so appealing in cognitivepsychology.
As the program absorbs more and more of the explanatory
burden, the physical factsunderlyingthe analyzing capacities be-
come less and less special to the analyzed system.This is why it is
plausible to suppose that the capacityof a personand of a machine
to solve a certain problem might have substantiallythe same ex-
planation,althoughit is not plausible to suppose thatthe capacities
of a synthesizerand of a bell to make similar sounds have sub-
stantiallysimilar explanations. There is no work to be done by
a sophisticatedhypothesisabout the organization of various ca-
pacities in the case of the bell. Conversely,the less Nveight
borne by
the program,the less point to analysis.At this end of the scale we
have cases like (10) in which the analyzed and analyzingcapacities
differlittleif at all in typeand sophistication.Here we could apply
the subsumptionstrategywithout significantloss, and thus talk
of functionsis comparativelystrainedand pointless.It must be ad-
mitted,however,that thereis no black-whitedistinctionhere, but
a case of more-or-less. As the role of organizationbecomes less and
less significant,the analytical strategybecomes less and less ap-
propriate,and talk of functionsmakes less and less sense. This may
be philosophicallydisappointing,but thereis no help forit.
NOOLREvEws 765
CONCLUSION
Almostwithoutexception,philosophicalaccountsof function-ascrib-
ingstatements and offunctionalexplanationhavebeencrippledby
adoptionof the assumptions (A) and (B). Though therehas been
wide-spread agreement thatextantaccountsare notsatisfactory,(A)
and (B) have escapedcriticalscrutiny,perhapsbecausetheywere
thoughtof as somehowsettingthe problemratherthen as part
of proffered solutions.Once the problemis properlydiagnosed,
however, it becomespossibleto givea moresatisfactory and more
illuminating accountin termsof theexplanatory strategy thatpro-
vides the motivation and formsthe contextof function-ascribing
statements,To ascribea functionto something is to ascibe a ca-
pacity to it which is singled out by its role in an analysisof some
capacityof a containingsystem. When a capacityof a containing
system is appropriately
explainedby analyzingit intoa numberof
othercapacitieswhoseprogrammed exerciseyieldsa manifestation
of the analyzedcapacity,the analyzingcapacitiesemergeas func-
tions.Sincetheappropriateness of thissortof explanatory
strategy
is a matterofdegree,so is theappropriatenessoffunction-ascribing
statements.
ROBERT CUMMINS
The JohnsHopkinsUniversity
BOOK REVIEWS
The Anatomyof the Soul. ANTHONY KENNY. Oxford:Basil Black-

well;New York:Barnes& Noble, 1973.ix, 147p. $8.50.
Not longbeforeI wasaskedto reviewthisbookI ordereda copyof
it fromthe catalogueof its Britishpublishers, Blackwell,on the
assumptionthat a new book of essayswould not, unlessclearly
indicatedin catalogueand advertisements, consistof unrevised
reprintsof alreadypublishedmaterial.But no: not onlyhave all
six of theseessaysappearedpreviously, but no fewerthan four
of themare on my shelvesin othercollections,threein cheap
paperbacks.There mustbe veryfewteachersof philosophy, and
even fewercollegeand university libraries,that do not already
own most,if not all, of thesematerials.It is naturalto ask,then,
whatspecialvalue thisbook can haveforitsintendedreadersand
buyers.
Well,thereis some new workhere,in theformof an eighteen-
page Appendix,consisting of an intendedencyclopedia articleon

The Journal of Philosophy Volume 72 Issue 20 1975 (Doi 10.2307/2024640) Cummins, Robert - Functional Analysis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Journal of Philosophy Volume 72 Issue 20 1975 (Doi 10.2307/2024640) Cummins, Robert - Functional Analysis

Uploaded by

Copyright:

Available Formats

Journal of Philosophy, Inc.

THE JOURNALOF PHILOSOPHY

tions are just that: assumptions.They have never been systemati-

(1) The heartbeatin vertebrateshas the functionof circulatingthe blood

He rejectsthe suggestionthat 'function'can simplybe replaced by

As Hempel sees the matter,themain problemwith thisanalysisis

Suppose,then,thatwe areinterested ofa

having actually been incorporatedinto the circulatorysystemsof

A less obvious counterexamplederivesfromthe well-attested fact

fromthe existenceand lengthof its shadow would fail to explain

x will performf in s which is causally relevant to x's presence in s, not x's

When we do this,we are elaborating on what we assume is the

tional characterizationin scientificexplanation, will see the prob-

(5) a. Vertebratesincorporating a beatingheartin theusualway(in the

Though by no means flawless,(5) has severalvirtues,not the least

These statementscan be understoodin eitherof two ways.(i) They

The second sortof use occurs as part of an account which,if we

them. Of course, if s is verydisadvantageousto organismshaving

6 In addition to the misunderstandingabout evolutionary theory just dis-

contained part, is that the systembe maintained in "adequate, or

analysisbreaksdown at some level forlack of functions,or perhaps

10 Even these applications have their problems. Frankfurtand Poole ["Func-

functioningof the sex organsresultsin the death of individualsof

functionas wings,as in the penguins or ostriches,would we cease

failsto giveany hintas to the pointof identifying certaineffects

1. Functionsand Dispositions.Something marbe capableofpump-

Thus, function-ascribing statementsimply disposition statements;

festationsas instancesof one or moregeneral laws, i.e., laws govern-

in a program or flow chart: each instruction (box) specifiesmanifestationof

strategy,17 for we often explain capacities by analyzing them. As-

by analyzing the organisminto a number of "systems"-the circu-

18 Indeed,what makessomething part of,e.g., the nervoussystemis thatits

3. Functions and Functional Analysis.In the contextof an appli-

(9) explains theintuitionbehind theregress-ridden (6): functional

4. Function-analyticalExplanation. If the account I have been

22 Of course, it might be thatthereare none but arbitrarydistinctionsto be

strategy.For my contentionwill be that the analytical strategyis

The Anatomyof the Soul. ANTHONY KENNY. Oxford:Basil Black-

You might also like