Species, Essences and the Names of Natural Kinds Author(s): T. E.

Wilkerson Reviewed work(s): Source: The Philosophical Quarterly, Vol. 43, No. 170 (Jan., 1993), pp. 1-19 Published by: Wiley for The Philosophical Quarterly Stable URL: http://www.jstor.org/stable/2219938 . Accessed: 01/03/2013 16:08
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Wiley and The Philosophical Quarterly are collaborating with JSTOR to digitize, preserve and extend access to The Philosophical Quarterly.

http://www.jstor.org

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

ThePhilosophical Vol. Quarterly 43 No. 170 ISSN 0031-8094 $2.00

1993 January

The Philosophical Quarterly

SPECIES, ESSENCES AND THE NAMES OF NATURAL KINDS
BY T. E. WILKERSON I. PUTNAM ON THE NAMES OF NATURAL KINDS At various timesin thehistory philosophy, in of and particularly recent therehave been attemptsto defenda doctrineof natural kinds. years, The central thoughtis that objects are membersof natural kinds in virtue of certain intrinsicproperties,quite independentlyof other objects and, in particular,quite independentlyof ourselvesand our of systems classification. Typical examples include kinds of chemical elements and compounds (carbon, oxygen, water, nitric acid) and kinds of biological individuals (chimpanzee, human, beech, oak). What makescarbon carbon,waterwater,a chimpanzeea chimpanzee, an oak an oak, is in each case a certainreal essence,a property setof or and sufficient membership thekind.We may for of properties necessary nothave thewit to discoverall thenaturalkindsthatexistbut theyare, as it were, out there,waiting to be discovered, and it is the job of scientists discoverthem. to in Presented thatway, thedoctrineofnaturalkindsis a metaphysical thesis,a view about what the world contains and how it works. It involves talkingabout natural kinds and real essences,and does not of involve asking, forexample, how names natural kinds functionin natural languages. Some philosopherswill consider that omission a veryseriousone, and will argue that a sane account of natural kinds of mustrest'uponan understanding theway in whichwe learn and use general names, and natural-kind names in particular. One very account ofgeneralnamescan be foundin a numberofpapers influential

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

T. E. WILKERSON

by Putnam,1and in thispaper I want to considerwhetherthe account applies to the names of biological kinds. So, although Putnam has a great deal to say about such names as 'gold' and 'water', I am only in interested what he has to say about such names as 'tiger'and 'lemon'. I shall discuss two alleged problems and, although I have broad with Putnam's view, I shall disagree over many important sympathy details. In thepapers thatformthefocusofour discussionPutnam develops the general outlinesof a view since dubbed 'externalism'- that is, the view that representations only if theyhave the rightsortof represent causal connectionwithobjectsofthekindrepresented. 'Meanings ain't in the head.' In particular,he argues, words get theirmeaning, not because we are in certainspecial internalstateswhen we use them,but ratherbecause theiruse is causally connected in an appropriate way with the objects referred When I use a word like 'tiger',I may well to. or have certainthoughts ideas runningthroughmy head, but it really does not matterverymuch what theyare. I may, forexample, have fierce Indian quadrupeds going through of feline, my thoughts striped, mind - a set of propertiesthat Locke would have called the 'nominal do essence'oftheanimalsin question- but such thoughts notdetermine the meaning the word 'tiger'. The crucial (though not perhaps the of only) considerationis that my use of the word 'tiger' should have the rightsortofcausal connectionwith tigers. How then are general names, and particularlynames of natural kinds,introduced? Accordingto Putnam,we are blessedwitha curious mixture of confidence and ignorance. We naturally find ourselves similarin colour, on focusing a group of objects that are superficially and we begin to think of shape, habit, geographical position, etc., certain objects as typical of the whole group. The general name is introduced originally as an indexical expression, that is, as an expression which allows us to point,literallyor metaphorically,to objects in the group. So, forexample, we focuson a group of animals fromIndia, and we introducethe that are large, striped,fiercefelines in word 'tiger' to point to any one of those, otherwords, to one of the Not only can we confidently apply the general name to typicaltigers. one of a numberof objects, but with equal confidencewe can list any India. from fierce felines of theproperties tigers: theyare large,striped, But - and thisis a crucialpoint- accordingto Putnam,theseproperties do not determinemembershipof the kind, and the list is not a list of
' See H. Putnam, 'Is Semantics Possible?', 'Explanation and Reference'and 'The Papers, Meaning of "Meaning" ', collected as Chapters 8, 11 and 12 of his Philosophical vol. II (Cambridge: Cambridge University Press,1975).

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES,ESSENCES,NAMES OF NATURALKINDS

of They merely defining properties tigers. help us, in Kripke's words,to 'fix the reference' the word 'tiger'.2Indeed, such propertiescould of not determinemembershipof the kind, because some members are abnormal and will lack one or more of them: thereare albino tigers, toothlesstigers,tigers that are mild and amiable, three-legged, or native to Regent's Park. At best, the properties define a tiger 'stereotype',and in practice our readiness to classify somethingas a tiger will depend on the extent to which it approximates to the stereotype. Confidencemay now give way to ignorance.We confidently apply theword 'tiger',literally metaphorically or pointingto certaintypical idea what actually determines tigers,but may not have the faintest of the membership thekind.In introducing name we suppose thatthere is someunderlying commonto all theanimalsin question,and property not merelyto the stereotypical animals, a propertythat, in nature, determinesmembershipof the kind. Moreover, in introducingthe word 'tiger',we have committedourselvesto applying it to anything that has that underlyingproperty,that has the relevant sameness relationto our originalstereotypical But we may not knowwhat tigers. the property and quite oftenwill not know what it is. At thispoint, is, Putnam argues, most of us are compelled to exploit the 'division of who labour'; thatis, we wait upon thediscoveriesofexperts, linguistic will reveal theunderlying thatsupportsour use of'tiger',and property all mutatis mutandis theotherunderlying thatsupportour use properties ofothernames fornatural kinds. II. THE FIRST PROBLEM: ORDINARY LANGUAGE Two implicitassumptions now need to be broughtto thesurface.First, Putnam never says explicitlyhow we are supposed to connect his account with a formaltaxonomy,but ifwe read between the lines, I thinkit is quite clear that he regardsthe names of biological kindsas names of species, and the underlyingpropertiesdeterminingnatural kinds as genetic propertiesthat determinespecies. Second, I do not thinkit would be at all unfairto characterize Putnam's account as his about thedivisionof Indeed, he introduces remarks quasi-historical. labour as a 'socio-linguistichypothesis'.3 He is therefore linguistic committedto an empirical theoryabout the way in which general
2 S. Kripke, Namingand Necessity (London: Basil Blackwell, 1980), passim,esp. pp. 134-9. 3 See 'The Meaning of "Meaning" ', p. 227.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

T. E. WILKERSON

names, are introducedand used in words,and especiallynatural-kind we naturallanguages such as English.If Putnam is right, should expect to to findthatEnglishnames ofbiological kindsnormallyrefer species, and that their use is supported by underlyinggenetic real essences whichdetermine of membership species.We should also expect to find thatwe wait upon theexpertsto tellus what thosegeneticreal essences are. At thispointwe encounterthefirst objection,namelythatordinary(hereafter OLC) do not fitPutnam's model. language classifications That is, Englishnames ofbiological kindsdo not pick out species,their use is not supported by genetic real essences which determine the species,and we do not wait upon the discoveriesof biological experts. counterJohn Dupre has produced a large number of interesting which I shall borrowwithgratitude, examples,4 mixingin a fewofmy own as we go. First,Putnam thinks biological names as pickingout of species,but many OLC names pick out kindsofa higherlevel than the species. For example, 'oak', 'beech', 'elm', 'willow', 'rose', 'iris', 'rhododendron','narcissus'pick out genera; 'duck', 'chrysanthemum', 'wren', 'woodpecker' pick out families;'gull' and 'tern' pick out sub'owl' and 'pigeon' pick out orders.Second, some OLC names families; failto pick out any formaltaxonomickindsat all: forexample, Dupre betweenprickly tellsus,Americanseasilydistinguish pears and chollas, but the distinctioncorrespondsto no taxonomic division. The same shrub, weed, perennial, point applies to the verygeneral OLC kindstree, and and vegetable fruit.Third, some OLC names collect verydifferent for unrelatedspecies together, example, 'wren', 'lily', 'daisy', 'orchid' and 'wolf'. Fourth,some OLC names have different uses in different dialects: according to Dupre, 'robin' collects one species in North in America,anotherin Britainand a genus offlycatchers Australia. What followsfromall this?I thinkit clearlyfollowsthat Putnam's account ofnatural-kind names cannot be regardedas a quasi-historical account of the developmentof our ordinary-languageclassifications. else follow? For example,do Dupr's exampleshave But does anything on generalattempts, such as myown, to defendan account any bearing ofnaturalkinds?Dupre believesthattheydo. For hiscarefulanalysisof OLC names is a central part of his argumentagainst the doctrineof natural kinds, and in favour of a view that he calls 'promiscuous realism'. Those who defend a doctrine of natural kinds claim that, between things, and differences although thereare many similarities
Review SeeJ. Dupre, 'Natural Kinds and Biological Taxa', ThePhilosophical (1981), passim. 4

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES,ESSENCES,NAMES OF NATURALKINDS

one set of similarities privileged,because theyare the real essences is which determinenatural kinds. In contrast,the promiscuousrealist is and each is important argues thatnone ofthesimilarities privileged, onlyrelativeto a particularconcern.As faras biologicalindividualsare concerned, a formal taxonomic classificationwill be of interestto biologists, horticulturaland agricultural classificationswill be of interest gardenersand farmers, to to gastronomicclassifications cooks and gourmets, taxidermal classifications taxidermists, to floralclassificand will ationsto florists, everydayclassifications servemanyofus very well for ordinary non-specialistpurposes. This view is a version of and differences that underpin each realism, because the similarities of are and theories, but system classification independentofour beliefs the realismis promiscuous,because therewill be an indefinitely large a number of possible systemsof classification,each one reflecting different interestand concern. Moreover, the various classifications may overlap and run acrosseach other:we would, forexample, expect to finda considerableoverlap betweenformal taxonomicclassifications and zoo keepers. (Dupre and the classifications gardeners,farmers of mighthave mentioneda shortstoryby R. K. Narayan, in which his gardenerhas only two kind names. One is an example of a name that I elsewhere5 have called a hybridname, and the otheris a veryhigh level natural-kind name: 'This, Sir,is a weed.' 'And that?''That, Sir,is a flowering plant.') I shall returnto promiscuousrealismlateron, and forthe momentI want to argue that Dupre's examples, though important and do underminePutnam's account, and are not interesting, not seriously to a doctrine of natural kinds. Let us return to counter-examples name by pointing, Putnam. He claims that I introducea natural-kind or metaphorically, apparentlytypicalexamples ofthekind, to literally that are associated with,but examples which share various properties do not strictly the kind name. And I suppose that the kind is define, determinedby some propertycommon both to the original stereoof typicalindividualsand to othermembers thekind,thoughI may not know what it is, and will usually need to call on expertsto tell me. Now thereis one important omissionin thisaccount, thoughPutnam The omissionconsists a in can remedyit withoutseriousinconsistency. failureto noticethatmany ofour attemptsto inventnew natural-kind names misfire, because our assumptionthat thereis some underlying commonto all themembers thesupposed kindmay be false. of property Our language has a long memory,and is litteredwith the debris of
5 See T. E. Wilkerson, 'Natural Kinds', (1988),p. 42. Philosophy

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

T. E. WILKERSON

numerousbotchedshotsat natural-kind For classification. example,we collectvarious stereotypical animals together, and use theirsuperficial features fixthe reference 'wolf' or 'wren' or 'lily' or 'orchid', not to of realizing that we are collecting different species together.Or our popular names forspecies suggesta close relationshipwhere none in factexists('ash' and 'mountain ash'; 'fox' and 'flying fox'). Or, in an elm, attemptto collect species,we collect genera (oak,beech, willow)or kindsthat correspondto no taxonomiccategories(prickly pear,cholla). At the limit,we even introducenatural-kindnames that turnout to referto nothingat all, such as 'animalcule', 'ether' or 'the Missing Link'.6 Such failuresare not surprising. Until fairly we recently, have had neitherthescientific nor thenecessarytechnology discover to expertise the real essences that determine physical, chemical and biological natural kinds. We have had to lean heavily, even exclusively,on on superficialfeatures, Lockean nominal essences,for theywere the our five only features accessible to our experimental instruments, senses,For example, in attemptingto classify plants and animals we have had to concentratesolely on gross morphological and physioand have inevitablybeen misled by them. I suppose logical features, of someone mighttryto argue that the survival antiquated systems of is and mightwonder why we pay so classification highlysignificant, littleattention thereceivedwisdomoftheexperts. to But again I cannot see thatsuch survivalis eithersurprising significant. ScottAltran or As has pointedout, even when a name has originally been introducedas a supposed natural-kindname (or, as he puts it, an 'underlyingtrait term'), whose use is underpinnedby some underlyingbut unknown it thatour suppositionis incorrect. property, may survivethediscovery in We may have such a stronginterest the objects in question that we are not in theleast concernedto modify classification our withthehelp ofexperts: the'phenomenologicalconcept' may persist anunderlying as trait termregardless science'sopinion on thematter.'Hawk' and of traittermsbecause theirusual 'sparrow' persistas underlying denotations are readily perceived to be componentsof local
If I read him correctly, case, wherethe Kripke denies thattherecan be thelimiting supposed natural-kindname turnsout to referto nothing at all. See his Namingand thatit is not the Necessity, I35n. Talking ofgold, he says: 'We knowin advance, a priori, p. case thattheitemsare typically fool'sgold.' But thatseemsan odd view. In a sensewe have or discoveredthatall our samplesare samplesoffool'sphlogiston, fool'sether!So we could not have knowna priori thattheyare not.
6

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES,ESSENCES,NAMESOF NATURALKINDS taxa.7

nature; the Tzeltal Maya, for instance, recognise similar Furthermore,even when we have the scientificexpertise and the we necessarytechnology, cannot guarantee that we shall produce the uncoverthe real essencesof things. answers,or routinely right So my response to Dupre's examples is to say that many of them at classificunsuccessful simplycommemorate attempts a natural-kind ation. The point,no doubt, tendsto get lostin Putnam's zealous flight from but intension, I thinkthathe can concede it,by slightly changing theemphasishereand there.Instead ofsayingthatI introducea name by pointing to a group of objects, which share some underlying of thatdetermines membership thekind,he should say thatI property to (sometimes may try introducethe name, and that I suppose wrongly) that the objects share some underlyingproperty. I conclude that Dupre's examples do not underminePutnam's account of biological names. III. THE SECOND PROBLEM: SPECIES, ESSENCES AND NATURAL KINDS We now turn to a second problem, which concerns Putnam's assumptionthat species are good examples of natural kinds,and his apparent confidence that membershipof species is determined by It underlying geneticproperties. mightseem obvious thathumans are humans,chimpanzeeschimpanzees,weepingwillowsweepingwillows, But and peculiar geneticconstitution. etc., in virtueoftheirdistinctive is thissuggestion open to a veryseriousobjection,namely thatspecies are not uniquelydetermined geneticconstitution. There is typically by enormousgeneticvariation between membersof the same biological considerablegeneticsimilarity betweenclosely kind,and thereis often related species. That is, genetic real essencesof natural kinds do not exist. Consider examples. Any gardener will be aware of the enormous varietiesofthe same species.No doubt the variationbetweendifferent but there can be explained by reference geneticfeatures, to variations be no geneticfeatures thatuniteall themembersofthespecies,all may There may noteven be a theindividualexamplesofthemanyvarieties. feature uniteall partsofthesame individual,for, to particularly genetic in plants, different parts of the same individual may be genetically
7 Scott Altran,'Ordinary Constraints the Semanticsof Living Kinds', Mind and on (1987), pp. 51-2. Language

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

T. E. WILKERSON

different. Gardenerswho noticegeneticvariationin a new stemcall it a cut it offand propagate it, in the hope that it will prove to be 'sport', commercially successful. Someone might suggest that the genetic of is feature thegeneticmaterialwe feature are lookingfor a structural for example, the number of chromosomespeculiar to each species. Despite the wide genetic variation between individual humans, we pairs of chromosomes.Sadly, mightsay, they all have twenty-three they do not! Down's syndrome,for example, is due to an extra in interested geneticsex and, as athleticsadministrators, chromosome, discoveredwithdismay,some humans carryan extrasex have testing, chromosome.In plants polyploidyis very common. In other words, because of an irregular divisionofgeneticmaterialin ancestors, many ratherthan thenormal have threeor moresetsofchromosomes, plants varietiesof the same species will have two. In some cases different different numbers of chromosomes: in apples, for example, a few varietiesare haploid, withseventeenchromosomes, many are diploid, some (such.as Bramley's Seedling) are triploid,with with thirty-four, and withsixty-eight, so on. some tetraploid, fifty-one, thatdetermine So themorewe attemptto isolatethegeneticfeatures themorehopelessthetaskbecomes. And perhapswe biological species, have unconsciouslybeen driven by a misleading picture of natural kindsas markedby nature,ratherin the way in which the boundaries between pigeonholes are clearly marked. Darwin, it might be said, demolished that picture for ever. It is not just that the work of evolutionarybiologistshas revealed many more boundaries than had and everbeen imagined- manymorespecies,manymoreintermediate is hybrid species - but the conception of boundaries a thoroughly misguidedone. If we are properlyto understandevolutionand genetic variation,and ifwe musthave some pictureto help us, thenthepicture of the mustsomehowreflect fluidity species,theway in whichtheymay into one another. Boundaries are no merge, almost imperceptibly, doubt helpfulforfieldbiologists, gardenersand zoo keepers,but they do not existin nature.John Dupre makes thissuggestion: in The existenceof species . . . may be seen as consisting the fact.If it were possible to map individual organisms following on a multi-dimensional quality space we would findnumerous willbe or clusters bumps. In somepartsofbiologytheseclusters almost entirely discrete. In other areas there will be a continuumof individuals between the peaks. It can then be thesepeaks.8 seen as the businessof biologyto identify
8 Dupr6, 'Natural Kinds and BiologicalTaxa', p. 82; also compare his 'Wilkersonon

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES, ESSENCES, NAMES OF NATURAL KINDS

The biologistStephenJay Gould also favoursthe pictureofcontinua: in Islands offormexist,to be sure: cats do not flowtogether a sea ofcontinuity, rathercome to us as lions,tigers, but lynxes, tabbies, and so forth.Still, although species may be discrete, theyhave no immutableessence.Variation is theraw material of evolutionary the change. It represents fundamentalreality of nature, not an accident about a created norm. . . . AntiWe essentialist forces to view theworlddifferently. us thinking mustaccept shadingsand continua as fundamental.9 Gould's readerswill noticethathe wishesto draw certainliberalmoral and political conclusionsbut, although those conclusionsare appealfor ing, theyare irrelevant mypurposeshere. I merelywish to explore and Dupre's, suggestion that species should be thought of as his, continua,ratherthan as separated like pigeonholes. forming Incidentally,we should note in passingthat theclaim that thereare of no geneticreal essences,and theclaim thatspeciesshould be thought as forming continua, are distinctclaims. The second claim probably That is, ifspecies are arranged in a continuum,it is implies the first. difficult argue that each species is determinedby a peculiar real to essence.But thefirst claim certainly does notimplythesecond. That is, ifthereare no geneticreal essences,it does not followthat thereare no discretespecies. Even if each species were discrete,and qualitatively quite distinct from every other, it would not follow that it was determinedby a single propertyor set of propertiesnecessaryand for of sufficient membership that species. Members ofdifferent species may turn out to be geneticallyvery similar; members of the same Again compare species may turnout to be geneticallyverydifferent. Dupre: and the existenceofdiscretespecies is one of the moststriking least disputableofbiological data. If one examinesthe treesor birds in a particulararea, it is apparent that thesefall into a fromone another in numerous number of classes that differ respects. But the essentialistconclusion that one might be this by temptedtodraw from factis dissipatedfirst morecareful characteristics study,which reveals that these distinguishing are by no means constantwithin the classes, and second by
Natural Kinds', Philosophy (1989), passim. 9 Stephen Jay Gould, The Flamingo'sSmile (Harmondsworth: Penguin, 1987), pp. 160-1.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

10

T. E. WILKERSON

in the extending scope oftheinvestigation bothspace and time, both ofintraspecific and the limitations similarity whereupon will difference become increasingly apparent.10 interspecific IV. A BAD REPLY: PROMISCUOUS REALISM So that is our second problem:ifnatural kindsare determinedby real essences,and if species are good examples of natural kinds,then we appear to have produced a contradiction, since species are not determinedby real essences.And realistsof a nervousand pessimistic dispositionmightfearthatrealismwill have to be givenup altogether, and that we shall have to endure some kind of anti-realism,some to we versionoftheview thatthefeatures attribute theworldare strictly of of ratherthanfeatures the of and our systems belief, features ourselves is But that pessimism too hasty.Even ifwe were world as it is in itself. eventuallyforcedto abandon essentialism(and I hope to salvage it lateron), therewould be at least one otherversionofrealismavailable, namelyDupre's 'promiscuousrealism',which I touched on earlier. In the course ofcriticizing Putnam, Dupre is struckby the range of Some names classifications. servedby our ordinary-language purposes are borrowed from biologists or are relics of discarded biological othersthoseofcooks the of some reflect interests farmers; classifications; or or and gourmets, timbermerchants, flower or arrangers, gardeners, is or or or furriers, hunters, zoo keepers,and so on. Each classification them. One classificadmirable forits purposes,and typicallyreflects uselessforotherpurposes: cooks and gardeners ation may be entirely between onions, garlic and ornamental insiston a strongdistinction Allium species,but biologistsdo not; zoo keepers' questionsabout the habits of animals cut rightacross taxidermists' questions about their for and suitability stuffing embalming; and so on. Each classification or but focuseson certain similaritiesand differences, no similarities or significantin themselves.A range of are privileged differences is similaritiesand differences significantonly relative to a special or of forexample, the interests farmers, gardeners,or cooks, interest, such as Putnam,who argues that etc. So, in contrastto the essentialist to of and differences interest biologistsare privileged,and similarities us somehow mark the real joints of nature,Dupre offers 'promiscuous realism': The realismderivesfrom factthatthereare manysameness the classes of organismsin ways relationsthat serve to distinguish
'1 Dupre, 'Natural Kinds and Biological Taxa', p. 89.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES,ESSENCES,NAMESOF NATURALKINDS

11

that are relevantto various concerns;the promiscuity derives fromthe factthat none of theserelationsis privileged." At therisk beingtediously of I thatpromiscuous repetitive, shouldstress realism is a version of realism.I say that, because so often in philosophical discussion the claim that some concept reflectsour our interests reflects quickly degeneratesinto the claim that it merely and anti-realism neverfarbehind. But that is not Dupre's is interests, on view. He wants to say that thefeatures fastened to by,forexample, farmers gardenersare realfeatures theworld. It isjust a factabout or of the world, not a factabout us, that some animals produce milk,that some plantssurvivetheBritish thatsome plantsare edible, and winter, so on. The only relevantfact about us is that if we are farmers, we will concentrate on one set of real features;if we are gardeners, we will concentrateon a ratherdifferent ifwe are zoo keepers,we set; will concentrateon anotherset; and so on. It is pleasing to be offered different of versions realism,but I want to fortwo main reasons.Both reasonsturnon rejectpromiscuousrealism, the general thought that the promiscuous realist leaves too much but away thatthereare manydifferent, unexplained.I concede straight often and classifications, thattypically overlapping, ordinary-language each classificationreflectsthe peculiar interestsof, say, farmersor or gardenersor cooks or gourmetsor flowerarrangersor taxidermists zoo keepers,or, of course,ofprofessional But my first biologists. worry concernsthereference theinterests to and purposesofeach group. The word 'interest'is ambiguous, and I suggest that Dupre's argument tradeson the ambiguity.Sometimeswhen we say thatsomeonehas an in interest something, mean merelythat theythinka lot about it, we devote theirleisurehoursto it, easilyfocustheirattentionon it, and so in on. The footballfan is interested football,the philatelistin stamps in botany. Sometimes,however,when we say that and the botanist in we someone has an interest something, mean that theyhave some in view. Despite the sad decline of the English word practical benefit 'disinterested',some of us still manage to mark the distinctionby failures of interest: someone may be distinguishingtwo different or disinterested. thelegal disputeis nothingto do with If uninterested, but me, but is fullofjuicy and scandalous detail, I will be interested If disinterested. my name is Jarndyce,and I am heartilysick of the whole thing,I may be uninterested have an interest. but
" Ibid.,p. 82.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

12

T. E. WILKERSON

in There are relatedambiguities suchwordsas 'anthropocentric' and which also occur in Dupre's comparison of 'anthropomorphic', classifications and scientific ordinary-language taxonomy(TC): The functionsof OLC, unsurprisingly enough, are overA group of organismsmay be whelminglyanthropocentric. distinguished in ordinary language . . . because it is economically or sociologically important . . . because its membersare intellectually and intriguing... furry empathetic ... orjust verynoticeable. avoids thisanthropocentric TC, hopefully, viewpoint.The numberofspeciesnames is hereintendedto reflect number the of species that exist. Nonetheless, even here there is an aspect. For an adequate taxonomy... must anthropomorphic also be practicallyusable.'2 sensesare being muddled together here. When I But several different I that a classification anthropocentric, might mean that it is is say that any analysis of the classification would logically anthropocentric, to involveessentialreference an actual or possibleperson.For example, our names forartefacts typically are a logically anthropocentric: table is witha flatsurfacecapable ofsupporting portablegoods the something of a normal human being; a house providesshelterforhuman beings; and so on. Narayan's gardener uses a classificationthat is, in part, for logicallyanthropocentric, a weed is a plant that human gardeners do when I say that a typically not want in the garden. But sometimes I is classification anthropocentric, mean that thereis a direct causal connection between the classificationand some practical human interestor concern. Taxidermists classifyskins and carcasses in a particular way because they want to find skins that survive preservationand stuffing; cooks are looking for thingsgood to eat; zoo keepers want to entertainvisitorsand preserve threatenedspecies; and so on. And finally, timbermerchants hope to sell wood at a profit; I is sometimeswhen I say that a classification anthropocentric, may mean very little. I may be tryingvainly to make somethingof the are obvious and drearyfact that classifications inventedand used by classifiers. I can now make myfirst pointagainstpromiscuousrealism.I suggest betweenmanyordinary-language thatthereis an important difference classificationsand scientifictaxonomies - between OLC and TC.
12

Ibid., pp. 80 and 81.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES, ESSENCES, NAMES OF NATURAL KINDS

13

connectedwithsome Much ofour workwithOLC is overwhelmingly purelypractical interest.(I say 'much' ratherthan 'all' because, as I explained earlier,some OLC names are the debris fromone or other failed attemptat TC.) It arises from,and is stimulatedby, a wish to a make money,or eat agreeable food,or satisfy desireto growthepretty and would make littlesense in the absence of such and the profitable, in Moreover,manyOLC names are anthropocentric practicalinterest. senses I distinguished.That is, they are the two most interesting to and implicitlymake essential reference logically anthropocentric, human desires('weed', 'houseplant','cattle', 'pet'); or theyare causally and theiruse causally depends on someone's desires anthropocentric, and practicalinterests ('shrub', 'groundcover', 'guard dog', 'poultry'); or, of course, both. In contrast, TC names have no necessary No doubt serious taxonomy connectionwith our practical interests. would grindto a halt iftherewereno one witha consuming professional would still or amateur passion forsuch things,but the classification recordtheimportant it have somepoint,for would, at leastin intention, distinctionsin nature. Furthermore,TC is neither logically nor to but essentially It implicitly causally anthropocentric. neitherrefers the desires and practical interestsof human beings, nor causally to Indeed, it is verydifficult say depends on thosepractical interests. be. Everyanswerseems of what the'practicalinterests' biologists might fatuouslycircular ('their practical interestis in biological classificthatis in ation'), trivial('theyare interested producinga classification or cynical to the point ofobstinateirrelevance('theywant to usable'), gain prizes and honours, astonish the crowd and humiliate their colleagues'). My second point against promiscuous realism is a point about concernis to explain causation and explanation. One centralscientific theworldaround us and to predicthow itwillbehave. But thatrequires an interest the causal powersof things.At the riskof hyperbole,we in is that the scientist concernedto expose the causal structure mightsay of reality.Now, as I have argued elsewhere,13 many of the kinds we do investigation, distinguish not lend themselvesto serious scientific connectedwiththecausal powersof because theyare notsystematically There can be no scienceoftables and chairs,as such,or even of things. trees,shrubs,perennials and annuals, as such, because the classificcausal powers. Of course, ations in question do not implicitly classify tables, chairs,trees,shrubs,perennialsand annuals obey causal laws, but only relative to another, quite different systemof classification, of namelyclassification naturalkinds.So thecausal powersofmytable
'Natural Kinds', esp. pp. 29-34. 13 See Wilkerson,

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

14

T. E. WILKERSON

are determinedby its being made ofcellulose,not by its being a table; thoseofmytreeby itsbeing an Acer, by itsbeing a tree;thoseofthe not perennialbyitsbeinga hosta,notbyitsbeinga perennial;and so on. So my second point against promiscuousrealism is that it overlooksan betweenOLCs and TC (or a formaltaxonomy): importantdifference thatimplicitly reveals typically, onlyTC is intendedas a classification thecausal powersofthings. And thatis whynaturalkinds,unlikemany of the kinds we find in OLC, lend themselvesto serious scientific investigation.Once again I make the obvious concession: cooks, farmers, gourmets,taxidermists, gardeners and other users of OLC have a profoundinterestin cause and effect.If not, they could not the guarantee that the meat was cooked, the skinpreserved, cattle fed and the seeds germinated.But my claim is that such guarantees are concernsof possibleonlyiftacitlywe move fromthe rathersuperficial OLC and lean heavilyon the causal storiessystematically interwoven withTC. V. A BETTER REPLY I therefore reject promiscuous realism, and continue to argue that, of available to us, among thevariousoverlappingsystems classification one (namely TC) is privileged,and the similarities and differences it records are privileged. But we are now faced once again with our originalproblem. Biological species are apparentlygood examples of natural kinds,and each natural kind is supposedlydeterminedby an intrinsicpropertyor set of properties necessary and sufficient for of the kind. But thereare typically intrinsic no membership properties or setsofproperties for of necessaryand sufficient membership species. The only candidate that looks remotelyplausible, genetic structure, fails to do the trick,because in the real world thereis considerable and intraspecific interspecific geneticsimilarity geneticvariation. We in the words of Gould and Dupre, thinkof species as forming should, continua, albeit with occasional bumps and clusters,rather than as forming quite separate,discreteislands ofform. In this section I shall offermy own solution of this problem, by exploitinga numberof importantpoints that have emergedfromthe discussionso far.One generalpointis thatan account ofnatural kinds arisesfroman interest causal explanation and prediction,and that in interest cannot be served by a promiscuousrealismwhich gives equal A weight to any usefulsystemof classification. second point is that, it is impossibleto findgeneticfeatures thatuniquely althoughtypically each species,therecan be no doubt thatan interest causal determine in

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES, ESSENCES, NAMES OF NATURAL KINDS

15

explanation inevitablyproduces an interestin genetic explanations. There is a striking betweenthe grossmacroscopicfeatures asymmetry of plants and animals and their microscopic- and especially their genetic - features.The macroscopic do not causally determinethe microscopic,but the microscopicdo causally determinethe macroscopic. Anyone seriouslyinterestedin explaining and predictingthe behaviourofplantsand animals would be foolish concentrate for to on, on of example, morphologicalfeatures, the structure leaf and root,of furand tooth. The fundamentalenginesof structure and change are genetic. A third intotheopen. Causal explanations pointcan now be brought are implicitlyor explicitlygeneral. I may be tryingto explain a but the particulareventor stateofaffairs, I am committedto offering same explanationofall relevantly similareventsor statesofaffairs. One common way of making the point is to appeal to the notion of a subjunctive conditional (though I am not clear what one says in languages thathave no subjunctive!).I may say, forexample, that the chloroticyellowing theleaves in this of plant was caused by therelative of thissoil, but I imply that any otherplant of a relevantly alkalinity similarkindwould if suffer from chlorosis it were growingin a relevantly similaralkaline soil. Even when theeventto be explained is historically unique, the explanation is implicitly general, and implies an appropriatesubjunctiveconditional. So to our problem. How can I continue to defend a doctrine of natural kinds, while conceding that species are not determinedby geneticreal essences?The simple (some mightsay, the wildlyheroic) answeris thatwe should abandon theassumptionthatspeciesare good examples of natural kinds.There are natural kinds,kindsdetermined in and sufficient membership each case, but for by properties necessary thekindsare typically not species. Indeed, in many cases the kind will be very much narrower than the species, and may have only one is member.My line ofthought as follows. us return a momentto Let for Gould's remarks about continua: We must accept shadings and continua as fundamental.... The taxonomicessentialist snailsin scoops up a handfuloffossil a singlespecies,triesto abstractan essence,and rateshis snails sees by their match to this average. The anti-essentialist in different hishand - a rangeofirreducible something entirely variation definingthe species, some variants more frequent than others,but all perfectly good snails.'4
14 Gould, The Smile, 161. Flamingo's p.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

16

T. E. WILKERSON

Now, ofwhat are the continua continua? Fairly obviously,we have a continuumofindividuals,wheretwo adjacent individualswould share but by manysimilarities would be distinguished at least one significant Even in a But clearly,we also have a continuumof kinds. difference. species characterized by considerable genetic variation, where some therewill individualsare geneticallyunique in all interesting respects, of be a continuumofkinds.For explanationofthefeatures each unique and thatrequiresthedependenceof individualrequiresgeneralization, certain featuresofa kindon certain other featuresof a kind.The real essence which is in part causally responsibleforthe behaviour of the geneticallyunique individual is the essence of the kindof which the unique individual is the only member. To sum up, then, my view is this. There are natural kinds. Each natural kind is determined by a real essence, a propertyor set of for propertiesnecessary and sufficient membershipof the kind in question. The real essence in turn grounds the causal powers of individual membersof the kind. Biological natural kindsare determined by genetic real essences which are causally responsiblefor the behaviour of individual members of the kind. But, since there is and intraspecific considerable interspecific genetic genetic similarity farmore biological natural kindsthan species. variation,thereare I should add threefurther comments- though I suspect that some readerswill regard the commentsas objections to my view. The first in commentis this.I have stressedthe role of geneticstructure causal and have argued that, even when explaining the explanations behaviourofan individual,I am implicitly explainingthebehaviourof individuals of a kind.I have also stressedthe range of interspecific and intraspecific geneticvariation. But presumably geneticsimilarity featuremay have a causal role and any geneticdifference any genetic in behaviour. a betweentwoindividualsmaydetermine difference their Since I want to say thatnaturalkindsare determined causal powers, by thattherewill be I am forcedto leave open theverydefinite possibility as many kindsas individuals. Is that not absurd? I thinknot. I would worriedifwe finished withas many individuals not be particularly up that thatitis veryunlikely as naturalkinds,but,fortworeasons,I think we shall. One reason is that many geneticfeaturesare causally inert. of carriedin strings DNA has of Only a small fraction the information role in affectingthe overall structureand behaviour of the any individual concerned.The second reason is that thereis a greatdeal of to the 'geneticfeature'.If I may help myself fogsurrounding expression the distinctionbetween functionand realization, which is normally context,the point can be put like this. prominentin a verydifferent

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES, ESSENCES, NAMES OF NATURAL KINDS

17

affect structure the and behaviour of When I say thatgeneticfeatures to theindividualconcerned,are thegeneticfeatures be characterizedin terms theirrealization of of (e.g., as complex moleculesconsisting atoms role of hydrogen,oxygen,carbon, etc.), or in termsof their functional of (e.g., in terms theiranatomical and physiological effects)? Certainly, ifwe characterizethem in termsof theirrealization,we may well be committedto as many kinds as individuals. But, if we characterize in of functional role,we can takeadvantage geneticfeatures terms their realizationsofthe ofthefactthattherewill be a wide range ofdifferent same geneticfeature, and in turnwill be able to regardtwoindividuals at different the level of realization as belonging to the same natural kind. Now thereis a compellingreason forcharacterizing geneticfeatures in termsof theirfunctionalrole. The crucial considerationis that the concept:thatis,a gene is precisely conceptofa gene is itselfafunctional the anatomical and physiological a structurethat directly affects features thewhole organism.Moreover,the causal powersofa gene of are the causal powers of the whole gene, not the causal powers of its in realization.As withany otherobject or structure which microscopic it does not matterwhat it function from we can distinguish realization, is made of,or how it is constructed, long as the causal powersof the as remainthe same. The pencilson mydesk,or whole object or structure at the washing machines in the shop, will be different the level of but at the functional level - qua pencils or washing realization, machines- will have the same causal powers. Similarly,theremay be realizationsof the same gene, many different arrangemany different mentsofatomswhichat thelevelofthewhole gene have exactlysimilar causal powers.So ifnatural biological kindsare determined genetic by is and ifgeneticstructure to be characterizedfunctionally, structure, thatis, in termsof the causal powersof the whole gene, we should not expect thereto be as many kindsas individuals. The second comment takes me back to Putnam. Even if my first to out to be true,I am stillcommitted farmorenatural commentturns kinds than species, for, typically,any species contains individuals affect one anotherin ways thatsignificantly different from genetically in are theirstructure and behaviour.But such differences not reflected our names fornatural kinds. Indeed, compared with the number of natural kinds,as I conceive ofthem,we have onlya handfulofnames. Nor is it worthour while to inventnames foreverynaturalkindeven if we are committedto developing an accurate taxonomy.As long as biologistshave the language of molecular biology, they can characthat underliekinds of terizethe many variationsin genetic structure

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

18

T. E. WILKERSON

all biologicalindividual,and willnotneed namesfor thekinds.But that undermines Putnam's account ofgeneralnames. Since, on my seriously account, both professional biologistsand ordinaryfolkwill draw the boundariesofnaturewitha verybroad brush,it is not truethatour use of general names is consistently underpinned by certain underlying of it properties the thingsin question,and afortiori is not truethat the extensionof the names is determinedby those underlying properties. Nor, of course, is it true that the extension of a general word is determined by the real featuresof objects, rather than by human decision.For, in introducing and usinggeneralnames,we mustdecide how closelywe wishour names to reflect fullrangeofnaturalkinds, the and how far we are prepared to extend our use of existingnames to individuals that share some genetic featuresof our original stereotypicalindividuals,but not others. a about the My thirdcommentis an attemptto confront difficulty the statusof formaltaxonomy.I began by entertaining thoughtthat everyspeciesis a naturalkind,determined a real essence,thatis,bya by or and sufficient membership for of property setofproperties necessary the kind. I went on to reject it, on the ground that species were not determined by real essences, and finishedby arguing that on my conceptionof natural kindstheremay be farmore natural kindsthan species. However, I consistently rejected any anti-realistor instrumentalistaccount of formal taxonomy: according to me, a formal taxonomy is supposed to tell us what kinds of thing really exist in nature.But I also resistedDupre's 'promiscuousrealism',and insisted that, although many systemsof classificationare available to us, a formaltaxonomyhas a special or privilegedstatus. Two connected questions immediatelyarise. First,am I guiltyof If inconsistency? I deny that species are typicallynatural kinds,am I not therebycommittedeither anti-realismor instrumentalism to to or realism?Second, what do I take taxonomists be doing? to promiscuous If theyare not markingout natural kinds,what are theydoing? The answersto thosequestionsemergeas soon as we pickout one important threadin my account of natural kinds.At the centreof my account is the notionofan intrinsic a thatdoes not consistin, property, property or depend on, a relation to somethingelse. For example, as I have argued in this paper, biological kinds are determined by intrinsic properties,for they are determinedby genetic properties.And the of in, geneticfeatures an organismlogicallydo notconsist or depend on, a relationto something else. In contrast, biologist'sconceptionofa the species is rather more complex. As I argued earlier, the interestin is intrinsic properties certainlythere,forthe biologistis concerned to

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions

SPECIES, ESSENCES, NAMES OF NATURAL KINDS

19

explain and predict the behaviour of plants and animals, and the fundamentalengines of structureand change are genetic. But the conception of a species also has at its centre the notion of a certain historicalrelation.That complexity interest of has, ofcourse,emerged comparatively recently. Until biologists became preoccupied with evolution, taxonomy was an Aristotelian attempt to discover the intrinsic substantialforms thatmarkedoff one kindfrom another.But, sincethepublicationof TheOrigin Species, have of particularly biologists become keenly interestedin descent. Their account of species is an account of stable populations of interbreedingindividuals, often geographicallyisolated fromother populations, and that account is meant to fitinto an overall historicalpictureof the descentofvarious othersby natural selection. populationsfrom I conclude that I am not guiltyof inconsistency. can deny that I species are typicallynatural kinds, but resistanti-realistor instrumentalist accounts offormaltaxonomy.My account of natural kinds, and the biologist'saccount of species, do ratherdifferent work. My is concerned exclusivelywith the intrinsic account of natural kinds and the causal powers,of individuals. In sharp contrast,the features, biologist's account of species rests both on considerations about and on considerations intrinsic about the historical geneticproperties relations betweendifferent ofindividuals.IfRichard Dawkins is groups right,'5the two sets of considerations are, of course, intimately connected: historicaldescent must be explained by natural selection, and the fundamentalmechanismof natural selectionis genetic.For it in consists theselectionand replicationofthe 'successful' genesand the a of the 'unsuccessful'. it is rightto interpret formal So disappearance forit is, at least in intention, ofan account taxonomyrealistically, part of real historicalrelations.And it is also rightto resistpromiscuous in realism,forthe biologist'sinterest the causal powers of things,the fundamentalmechanismsof change, is enough to justifythe special statusofformaltaxonomies.'6 UniversityNottingham of

15 See R. Dawkins, The Selfish Gene(Oxford:OxfordUniversity Press,1976 and 1989) and TheExtended Press,1982). Phenotype (Oxford:OxfordUniversity 16 I am grateful comments an earlierdraft from for on GregMcCulloch and thereferees forthis journal.

This content downloaded on Fri, 1 Mar 2013 16:08:30 PM All use subject to JSTOR Terms and Conditions