IAN HACKING

ON BOYD

(Received14 March,1990)

Boyd's "essay" (as he rightly designates it) develops many realist

themes.I take up the three that bear on my work.I welcome his theory
of clusters,and contrastit with the idea of familyresemblances.I query
his history.I resistthe incursionof naturalkindsinto humankinds.

1. CLUSTERS AND FAMILY RESEMBLANCE

I warmly welcome Boyd's theory of homeostatic property clusters,

sketchedon pages 15-18 of his reference(1989). It is the best recent
contributionto the doctrineof naturalkinds.Had it been in printwhen
I wrote or even revised my paper, I would have noted how much it
would have pleased Russell and most other writers in my tradition.
Russellsaid cheerfullythat naturalkindsare like topologicalneighbour-
hoods. Boyd's homeostaticpropertyclusters are a naturalexplanation
of this idea. It has many meritsfor which Russell and Mill did not even
grasp. Arguablyit provides Mill with a defense againstPeirce's most
salientcriticism,and allows Mill to retainthe idea of "realKinds"with
an almost inexhaustiblenumberof propertiesthat do not intrinsically
follow one from the other. It also fits readily into many related
approaches,for example, the "resemblance"version of the doctrine
(which I do not admire) best stated by H. H. Price. It enables us to
overcome Goodman's problem of imperfect communitythat usually
besets resemblanceaccounts.
Boyd notes above and in (1989) that names for biological species
have been prime candidates for natural kind terms. They are his
favouredexamplesin illustratingand vindicatinghis theory of homeo-
static property clusters. That is why his important contribution is
irrelevantto familyresemblances.He wronglyregards"propertycluster
term" and "familyresemblance word" as meaning much the same.

PhilosophicalStudies61: 149-154, 1991.

? 1991 KluwerAcademicPublishers.Printedin the Netherlands.

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions
150 IAN HACKING

"Familyresemblance"has quite other connotations,with Wittgenstein

setting the pace. Wittgensteinwas deliberatelyopposing family resem-
blance nouns to nouns that name,for example,biologicalspecies. In his
(1989) Boyd takes some pleasure in refuting unnamed "ordinary
languagephilosophers"who take the characteristicsof clusterterms to
be conceptual ratherthan causal. Let us be clear that whoever those
peoplewere,Wittgensteinwasnot amongthem.
I supplementedWittgenstein,perhaps mistakenly,by suggestinga
connectionbetween familyresemblanceand social kinds.Boyd exacer-
bates any errorby quotingme out of context,as sayingthat such "kinds
are constructedalong the lines of family resemblance,and what puts
thingsin to a familyis not naturebut people in concert."That sentence
was one of two theses (a) and (b) that I mentionedbut did not asserton
p. 116. I attributedthem, a little maliciously,to George Lakoff.Later,
whatI myselfsaidwas,

I conjecturethat a great many family resemblancenouns collect the objects to which

they apply in a "non-natural
way"- this is, they rely on social factorsand may properly
be calledsocialkinds(p. 123).

Thus I must insist on (i) my words "conjecture"and "agreatmany",and

(ii) on the fact that Boyd'spropertyclustertermsare not whatWittgen-
stein meant by family resemblancenouns. Since I also hold that there
are many differenttypes of familyresemblances,I can have no serious
objection to includingBoyd's clustersin the extended familyof family
resemblances,once one sees (on anotheroccasion) how extended and
diverse that familyis. If we did include Boyd's clustersthen, platitudi-
nously,my "agreatmany"abovewouldstillremaincorrect.

2. WHEN DID THE TRADITION BEGIN?

Boyd's "onlyslightlyanachronistic"strategyis to move the traditionof

naturalkinds back to Locke, to identify an intrinsictension between
Enlightenmentnominalismand predictingthe future,and then to trace
the evolution of this difficultyinto the present througha quick survey
of verificationistphilosophy, finally concluding that only a sturdy
realism about natural kinds will save the day. UnfortunatelyI don't
recognize much in Boyd's reconstructedhistory, which mentions only

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions
 ON BOYD 151

one figure in my tradition,and which describesrecent philosophyin a

waythatis unfamiliarto me.
This is not the occasion for an historiographicsermon,especiallyas I
provided no history in my paper. I must say that Locke is not part of
the traditionof naturalkinds. It is true of course that with any of the
enduring themes in Western thought - unity, atomism, continuity,
substance,universalsor what have you - we can reach back as far as
we like to find precursors.The traditionof naturalkinds has simulacra
in Locke, in Buridan,in Aristotle,in Heraclitus.But the traditionitself,
with the concerns with induction that I noted, and its conception of
kinds as causal or historicalentities, could not have come into being
before about 1800, and is a minor element in a very majorredistribu-
tion of ideas. One item in that largerredistributionthat helped trigger
the concern for kinds is the transformationof natural history into
biology (a word invented about 1801 for a new type of knowledge).
Locke lived in an epoch when representationwas king, and when a
classificationwas intended to representin a table the surfaceand still
features of form or function of a preservedexemplar.It was ideas (a
concept almost inaccessibleto us) that were the workmanshipof the
understanding,not kinds. The question was always, does an idea
representits object? It was Locke's view that there was nothingin the
objects from which we formed general ideas that could determinethe
boundaries between ideas, so that those boundaries had to be the
workmanshipof the understanding."I would not here be thought to
forget"- so he begins the "workmanship of the understanding"
passage
- "muchless to deny, that Nature,in the productionof things,makes
several of them alike: there is nothing more obvious .. .". There were
resemblancesin Nature, but no manifest boundariesfrom which the
ideas in the understandingmight be formed and which they could
represent. Michael Ayers has an excellent account of some of the
reasons why Locke was so obsessed with boundaries,and his account
can be augmented.' This problematic has no connection with the
traditionof naturalkinds. None of my authorsalludes to Locke. Mill
introduced,and the traditioncontinuedto use, a new word, "kind",for
a new discussion. None of my authors was interested in boundaries
between kinds. Boyd betrays a misapprehensionhere. In his first
paragraphhe writesthat in the tradition,a kind is "(atleast on a certain

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions
152 IAN HACKING

idealization) defined by a set of necessary and sufficient properties

(relations etc.) .. .". I did not mention NASC nor did any of my
authors,in this connection.(Perhapshe was surprisedthat I, and those
for whom I speak, welcome his clusterconcept so warmly:we welcome
it because we never cared a fig for NASC). We would have to embark
on some historyto show why the doctrineof naturalkinds commenced
when it did, but that it beganin the earlypart of the nineteenthcentury
is incontrovertible.Thus would I evade Boyd's stratagemof linking
handswithLocke.

3. SOCIAL CONSTRUCTION

Boyd's concludingremarksaddress a numberof views that Boyd may

believe me to hold but which are not stated in my paper.I mentioned
social kinds and humankinds, as relevantkinds that interestme. In the
paper I venturedno analysis.Since Boyd criticizesviews about social
kindsthathe thinksI hold, I mustnow verybrieflyturnto thattopic.
First,I find repugnantBoyd's suggestionthat such kinds are "inscru-
table",a word he uses repeatedly,as often as four times to a paragraph.
I read with astonishmentabout "Hacking'sinsistencethat the adoption
of definitionsfor social kinds has a real but inscrutableinfluenceon the
propertiesof the social objects under study."I have never insisted on
any such thing.I have never spoken of real but inscrutableinfluences.I
have almost never considered "definitions"for social kinds, definition
seldombeinga pertinentconceptin thisarea.
Far from mutteringabout inscrutabilityI painstakinglyscrutinize
some kinds of people and their behaviour- "multiplepersonalities",
"childabuse",even the metaconceptof "normalcy".Many such kinds
are close to some moral kinds, and are immenselyinfluentialin day to
day moral reasoning.Boyd imagines that I am into "social construc-
tion."In fact I do not hold such kinds to be (socially)constructedin any
careful sense of that phrase. I resist terms like "social construction"
because their very latinismreeks of false science.I preferplain English
metaphorssuch as "makingup people", and I do think that there are
senses in which many of our traitsof characterand types of action are
made up.2That is to say, in part, that the historiesof humankinds are
quite differentfrom those of naturalkinds, for as some classifications

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions
 ON BOYD 153

are formed and moulded they loop back, interactwith, and alter the
individualsand the types of behaviourto whichthey apply.At a deeper
level, they help determinethe very space of possibilityof action.This is
hardly surprisingif all actions are actions under a description, and
descriptionsinteract with agents. They affect not only who I am but
also my projects,the kind of person that I mighthope to be, to value,to
trust or to love. Yet as we change to fit the kinds, so we change the
kinds to better sort those who fall under them because the individuals
- often ourselves - change in the light of being described. There is
nothing "inscrutable"about this, although careful scrutiny is very
demanding.
Boyd understandssome of this idea, althoughit is not in my paper,
and holds that "theinfluenceof classificatorypracticeson causal struc-
ture alwayssuperveneson ordinarycausalmechanisms."Superveneson
ordinarycausal mechanisms!Does that have any meaningat all? Boyd
takes "fish fork" to be purely conventional, but I assume that the
interactionof the classification"fishfork"with the dining practicesof
the elite and the makers of expensive silverware "supervenes on
ordinarycausal mechanisms"- if the phrase has any meaning.So I
would not be one to deny (if it made sense) that classificatorypractices
havethishappyknackof supervening.But so what?
It is preciselysuch abstracttalk that my own studiesof humankinds
try to escape - and which we can escape if we stop blindly following
the pattern of natural kinds. Boyd writes that "barringa successful
defense of social constructivism,the project of extendingthe theory of
naturalkinds to historicalkindsis secure."Thatis exactlythe move that
I resent. To repeat a sentence from the last paragraphof my paper,
"Thoseof us who care about other relevantkinds need not be bullied
into sayingthat they are, or are not, just like naturalkinds."I need not
mount "a successful defense of social constructivism"to block Boyd's
"project".Me? Shoulda notorious(experimental)scientificrealistabout
the unobservedentitiesused in most of the naturalsciencesmount such
a defense? In my paper I urged that there are many modest but
excellent insightsin the great traditionof naturalkinds - hoping then
to move on to other relevantkinds, withoutbeing accused of mindless
all-purpose constructivism.What is interesting about some other
relevantkinds is understandinghow they work on us and how we work

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions
154 IAN HACKING

on them to form the very possibilitiesthat are open to us as people.

Many human kinds have powers unknownto naturalkinds. They are
instrumentsand agentsof power and"knowledge, but also of caringand
of stewardship.Some wrongly think we can escape their force into
some gloriousfreedomto be ourselves,but in fact they are our essential
partners,withoutwhichwe are mereflesh andnerve.
Thus I think that the role of human kinds in our lives, and in the
human and social sciences too, has little to do with those spectator
sports so admiredin some theories of naturalkinds, namelyinduction
and explanation.And returningfrom humanto naturalkinds, it will be
noticed in my paperthatit was not inducingand explainingthat seemed
to me the hallmarkof naturalkinds,but ratherdoing and using,melting
and breeding,miningand cultivating.That vision is more reminiscent,
for those who wouldreminisce,of Aristotlethanof Locke.

NOTES

Michael S. Ayers, "Locke versus Aristotle on Natural Kinds," The Journal of

Philosophy78 (1981), 247-272.
2 The programmaticstatement, with
some examples, is lan Hacking, "Makingup
people",in T. C. Heller et al. (eds.), ReconstructingIndividualism:Autonomy,Individ-
uality and the Self in WesternThought,Stanford:Stanford University Press, 1986,
222-36.

InstituteforHistoryand Philosophyof Scienceand Technology

Universityof Toronto
Toronto,OntarioM5SIK7
Canada

This content downloaded from 91.229.248.157 on Tue, 10 Jun 2014 14:17:39 PM

All use subject to JSTOR Terms and Conditions