Passing, Traveling and Reality: Social Constructionism and the Metaphysics of Race

Author(s): Ron Mallon

Source: Noûs, Vol. 38, No. 4 (Dec., 2004), pp. 644-673
Published by: Wiley
Stable URL: https://www.jstor.org/stable/3506217
Accessed: 31-05-2019 00:41 UTC

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].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Wiley is collaborating with JSTOR to digitize, preserve and extend access to Noûs

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 NOUS 38:4 (2004) 644-673

Passing, Traveling and Reality: Social

Constructionism and the Metaphysics of Race
RON MALLON

University of Utah

Among race theorists, the view that race is a social construction is widespread
While the term 'social construction' is sometimes intended to mean merely that
race does not (as once believed) constitute a robust, biological natural kind, it
often labels the stronger position that race is real, but not a biological kind.
For example, Charles Mills (1998) writes that, "the task of those working on
race is to put race in quotes, 'race', while still insisting that nevertheless, it exist
(and moves people)"(xiv, italics his). It is to "make a plausible social ontology
neither essentialist, innate, nor transhistorical, but real enough for all that"
(xiv). Racial constructionism, thus conceived, is a metaphysical position that
contrasts both with the view that race is an important biological kind (racial
naturalism) and with the more recent claim that race does not exist (racia
skepticism). The desire for a constructionist metaphysics of race emerge
against the background of a cluster of normative disputes, including:

1. Labeling Practices: Is the use of any racial terms to pick out various
human groups or subgroups by arbitrary bodily features useful or
permissible?
2. Terminology: Should term x be used in social life, social theory or
social science?

3. Significance of Racial Identity: What is the value of racial identifica-

tion of oneself or others? Is racial identification morally significant?
Is the social enforcement of racial identification morally permissible?
Is it morally required?

To simplify, we can characterize these normative disputes as disputes o

the value of 'race' talk.' By "'race' talk" I mean talk that uses 'race' an

? 2004 Blackwell Publishing Inc., 350 Main Street, Malden, MA 02148, USA,
and P.O. Box 1354, 9600 Garsington Road, Oxford, OX4 2DQ, UK
644

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 645

other race terms including terms such as 'black', 'white', 'Asian', etc. to
classify people (including oneself) or differentially treat them. A rough
characterization of these normative disputes has it that at one pole of
these debates short-term eliminativists want to eliminate 'race' talk quickly
(e.g. Appiah 1995, D'Souza 1996, Muir 1993, Webster 1992, Zack 1993). At
the opposite extreme, long-term conservationists hold that racial identities
and communities are beneficial and that 'race' talk-suitably reformed from
the excesses of racism-is essential to fostering them (e.g. Outlaw 1990,
1995, 1996). In between these two extremes, there are many who believe that
race talk is necessary (and perhaps inevitable) in at least some domains in
the short term because of the pervasive existence of racial division and the
effects of such division in modern life but who differ with regard to its long
term value.
Normative disputes give rise to a concern with the metaphysics of race
because of the role metaphysical arguments play in supporting normative
conclusions. For example, Naomi Zack argues that "the ordinary concept
of race in the United States has no scientific basis" (1993, 18), and
K. Anthony Appiah writes that, "the truth is that there are no races: there
is nothing in the world that can do all we ask 'race' to do for us" (1995, 75).
According to Zack and Appiah, 'race' talk makes reference to a set of racial
properties that literally do not exist, and, for each, this provides a reason to
eliminate such talk as mistaken.2 According to this line of argument, the
correct metaphysical position (racial skepticism) provides a reason to
endorse a particular answer to the normative question (eliminativism).
Like their eliminativist opponents, short and long-term conservationists
about 'race' talk argue from a metaphysical (constructionist) account of
race to conclusions about the need for 'race' talk. We can see this appeal in
Lucius Outlaw's claim that, "For most of us that there are different races of
people is one of the most obvious features of our social worlds" (1990, 58),
as well as in Mills's insistence that race "exists" and "moves people." Such
theorists argue that theories or policies that do not make reference to race
leave something out (e.g. Outlaw 1995, Mills 1998, Omi and Winant 1986,
1994, Root 2000, Sundstrom 2002).
But what is this thing? Constructionist theorists are loath to embrace
racial skepticism, but they (like racial skeptics) wish to avoid a commit-
ment to racial naturalism. Instead, constructionists hope to chart a third
metaphysical option, one that holds that race exists, but as a product of
particular social practices. But what, exactly, does it mean for race to be
socially constructed? In recent years, a variety of philosophers including
Robert Gooding-Williams (1998), Mills, Adrian Piper (1992), Michael
Root (2000), and Iris Marion Young (1989) have turned their attention
to this metaphysical question.3 In what follows, I argue that despite the
progress these accounts represent, they nonetheless fail to arrive at an
adequate constructionist account of race. The reason, I suggest, is that

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
646 NO US

constructionists are committed to three mutually unsatisfiable c

on an acceptable account of race, and no univocal account can s
three. Faced with this failure, one might be tempted to abandon
tionism for racial skepticism. Another alternative is to abando
another of the proposed constraints in favor of an attenuated c
tionism. I argue, however, that we need not choose between th
natives. Once we set aside the worry about which account is app
considered an account of race, we see that skepticism and the
attenuated forms of constructionism are best understood not as meta-
physical rivals at all. Rather, these positions share a broad base of m
physical agreement.
Here's how I proceed. In Section 1, I briefly discuss the wide
endorsed view that race is not a biological natural kind. Then, in Se
2, I focus on whether constructionists can account for the phenom
passing. Passing occurs whenever a member of some category is per
(and allows herself to be perceived) as a member of another, mutu
exclusive category, for example a white person passing as black
black person passing as white. Walter Benn Michaels (1994) charge
constructionist accounts of race cannot account for passing. I show
constructionist account of race drawing on work by Gooding-Will
(1998), Mills (1998), and Piper (1992)-a version of what I call a t
account of race-can answer Michaels's critique. In Section 3, I turn
discuss the claim by Root and others that race 'does not travel' bey
particular cultural-historical site, and I will show that constructionis
make sense of such claims by endorsing an interactive kind accoun
race. Constructionists thus have ready answers for the challenges rai
passing and not traveling. Unfortunately, the two answers invoke
different accounts of what race is and thus do not provide us w
univocal account of race. For this reason, I consider a third, instituti
account of race. Such an account, I suggest, can accommodate b
passing and not traveling. But, in Section 4, I argue that institutio
accounts fail to meet a third condition of adequacy on a construct
account of race: accounting for the reality of race. I conclude that
single constructionist account of race can accommodate all the theo
needs to which constructionists wish to put it. In Section 5, I argue
the failure to find a univocal constructionist account leads us to a v
of alternative accounts of race that abandon one or more of the ade
constraints I have suggested. But rather than choose between
accounts, I argue the divisions among them are not metaphysically
nificant. In confronting racial phenomena, skepticism and the varie
constructionism share a broad base of metaphysical agreement, an
adequate racial theory should exploit this agreement and work to d
guish all the features of racial phenomena that matter.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 647

1. The Social Construction of Race and the Ontological Consensus

K. Anthony Appiah uses "racialism" for the view that,

we could divide human beings into a small number of groups, called "races," in
such a way that the members of these groups shared certain fundamental,
heritable, physical, moral, intellectual, and cultural characteristics with one
another that they did not share with members of any other race.4

Let's draw a distinction between thin and thick racial endowments. Thin
racial endowments are thin clusters of properties that include one's g
type, and more or less genetically determined phenotypic features suc
skin color, eye color, and body morphology that have traditionally
associated with racial categories. Include also in one's thin racial endo
ment various relational properties including one's ancestry that are extrin
to the individual, but are heritable." It is on the basis of perceptible eleme
of my thin racial endowment that people everyday ascribe me to a ra
category. In contrast, a thick racial endowment is what racialism is co
mitted to: the existence of thick clusters of properties including th
properties in the thin racial endowment and, in addition, a set of "physica
moral, intellectual, and cultural characteristics" that are held to be
explained by one's biological endowment. Racialists thus hold that race is
an important biological and social category. If racialism were correct,
knowing someone's race would provide you good reason to believe other
things about the person's physical, moral, intellectual, and cultural char-
acteristics. However, there is now widespread agreement among scientists
and social theorists alike that racialism is wrong since there are not the sorts
of thick clusters of biologically determined human traits envisaged by
racialism.6 Appiah provides a succinct explanation:

while there are some characteristics that we are very good at recognizing-skin
color, hair, skull shape-that are very unevenly geographically distributed, the
groups produced by these assignments do not cluster much for other character-
istics. (1996, 68)

Because genotypic or other biological racial differences do not determine or

explain thick clusters of phenotypic racial differences, racialism is false. I
will call the denial of racialism the ontological consensus.7 Racialism is a
moral and political concern since the presence of thick clusters of biologic-
ally determined character traits may provide a foundation for racist claims
of racial moral, intellectual or cultural superiority. I will follow the practice
of saying that racist doctrines involve both the endorsement of racialism
and additional claims of the superiority of one or another race.8 The
ontological consensus undermines racism by undermining racialism.9

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
648 NOUS

Any constructionist theory of race should be consistent with t

consensus, but notice that the ontological consensus leaves i
questions such as "does race exist?" and "ought we to use 'ra
could, for example, reply to the former question by saying tha
like the racial kinds posited by racialist biology exists, race doe
we could conclude that races do exist, but that racialists were a
mistaken about what races are. Perhaps we should say that
correctly differentiate people on the basis of thin racial endow
explanatory power of such concepts is extremely limited. Alter
could say that races did not exist in the past, but now they do
social construction. These options will become clearer as we g

2. On Passing

Passing is an important phenomenon in the history and experie

America, and it is also discussed by a variety of construct
theorists." Passing is problematic for constructionists, sin
involve a person objectively belonging to one race while bei
belong to another. Since constructionist accounts of race ai
account of what race is, these accounts ought also to allow us to
passing. On the other hand, if a constructionist account cann
of passing, then such accounts must be inadequate to unde
complexity of race. Thus,

The Passing Constraint: On a constructionist theory of ra

should be possible and explicable.

Michaels (1994) charges that constructionist accounts of rac

cisely these grounds. In this section, I will set out Michaels's
will draw upon and extend recent constructionist accounts
Williams and Mills to answer the charge. I will then consider th
of race that underlies this response.

2.1 Michaels's Critique of Constructionism

Michaels's critique stems from his general view that "for the id
identity to do any work beyond describing the beliefs peopl
and the things they actually do, it must resort to some v
essentialism it begins by repudiating."'2 Passing provides a
test case, Michaels believes, because "the very idea of passin
takes the form of looking like you belong to a different race o
you belong to a different race-requires an understanding of
thing separate from the way you look and the way you act"
continues: "If... to see race as a social construction is inevi

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 649

unwillingly and unknowingly) to essentialize it, then race really is either an

essence or an illusion" (1994, 769). Michaels thinks the choice is clear: either
hold on to racial ascription and be committed to some sort of pernicious
essentialism, or abandon it and abandon the idea that race exists. A racial
constructionist middle ground does not exist.
Michaels's argument takes the form of a dilemma, the first horn of which
can be stated in a simple and seemingly compelling manner:

II. Social constructionists hold that "race [is] nothing but culture"
(1994, 768).
12. Culture is "nothing more than what we do and believe" (1992, 682),
"a distinctive array of believes and practices" (1994, 768).
13. So, race is nothing more than what we do and believe.
14. Therefore, "to believe and practice what the members of any race
believed and practiced would, by definition, make you a member of
that race" (1994, 768).
15. This makes one's race a voluntary choice, for people could change
their racial identity, siblings could belong to different races, people
who were as genetically unlike each other as it's possible for two
humans to be could nonetheless belong to the same race. None of
these things is possible in the U.S. today. And, were they to become
possible, we would think not that we had finally succeeded in
developing an antiessentialist account of race but that we had
given up the idea of race altogether. (1994, 768)

To accept 11-14 is to accept an identification account of race. On this

account, one's race is determined simply by what one believes and how
one acts.13 Michaels's argument shows that an identification account of race
makes passing impossible.
Robert Gooding-Williams (1998) challenges Michaels's argument by
denying that (14) follows from (13). The argument assumes that if race is
"nothing but culture" that means that what makes someone a member of a
race is the racially differentiated way they act. But race can be "nothing but
culture" and not depend on the individual beliefs, actions, or identifications
of those racially labeled. Rather, as Gooding-Williams points out, construc-
tionists may hold that race is "nothing but culture" in that practices of
racial classification are nothing but culture, and that it is these practices that
determine the criteria for racial membership (21).
Michaels considers this possibility, but he thinks that it falls upon the
other horn of his dilemma, and so it simply displaces the problem. The
argument for this other horn is again straightforward:

C1. Alternatively, constructionists hold that one's race is a matter of

how one is classified.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
650 NOUS

C2. This makes it the case that "if you are perceived as b
black" (1994, 767-768).
C3. And this makes passing impossible, since to pass as a
race R would be sufficient to be a member of race R.

Gooding-Williams thinks that Michaels's argument here stems from a con

fusion about the position he attacks-Adrian Piper's (1992) discussion of
passing. Michaels's critique is directed at Piper's claim that "what make
blacks black is 'the shared experience of being visually or cognitively ide
tified as black by a white racist society, and the punitive and damagin
effects of that identification'."14 However, as Gooding-Williams point
out, Piper's account has more than merely visual modes of identification
to rely on:

Where Piper sees the American practice of racial classification as incorporating

both visual and cognitive identifications, Michaels pays attention only to visual
identifications, that is to the perception of individuals as black .... Piper's
reference to cognitive identification is meant, I assume, to flag the fact that
the American practice of racial classification involves criteria entailing that
someone perceived to be white can be black and that someone perceived to
be black can be white. (21)

Gooding-Williams is correct that a constructionist account may draw on

more than perceptual criteria in determining racial ascription. Ordinary
racial ascription draws a distinction between looking like a member of a
race-what we might call apparent race-and actually being a member of
that race.
But despite Gooding-Williams's defense, Piper's account cannot answer
Michaels's objection. For Michaels can restate his objection without refer-
ence to perceptual identification, which he does:

The point of [Piper's] definition is that being black means being identified by a
white racist society as black. On what grounds, then, can someone who is not
identified by that society as black be said to be black? (1994, 768)

The problem with Piper's account is not one that can be resolved only by
appeal to cognitive criteria. The problem is that Piper's conception of black
identity relies not on apparent race (as Michaels initially implied) nor on
how a person would be classified under some set of cognitive criteria, but on
the set of actual experiences shared by people in virtue of their being subject
to a common classificatory system. Call this an experiential account of race.
On an experiential account of race, anyone who escapes actually being
classified as a member of race R (and thereby escapes the common experi-
ences of R's) is not an R.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 651

Experiential accounts of race are important, but they cannot account for
the depth of some ordinary intuitions about race and passing. The pressing
problem for meeting the Passing Constraint is that paradigm cases of
passing suggest that at least one of the ordinary notions of race has it that
someone could racially pass even to herself, and even throughout much of her
life. As a literary example, consider Mark Twain's (1976) The Tragedy of
Pudd'nhead Wilson, which details the story of two indistinguishable infants,
one a black slave and the other a white heir, whose positions are switched.15
The novel climaxes with the restoration of their racial positions, and the
narrative depends in part on the idea that the children were passing as
members of another race-not in virtue of their racial experiences, but
only in virtue of their parentage and ancestry. Real life cases that suggest
a similar nonexperiential notion of race at work include both the infamous
case of Susie Phipps, a woman declared black by a Louisiana court despite
always having thought she was white, and the case of Gregory Williams who
'discovered' that he was black only as an adolescent when his parents
divorced.16 An experiential account of race would allow for passing only
in the cases in which a person has experienced being classified as a member
of one category, but chooses for a time to pass as a member of another.
Insofar as the ordinary notion of race and passing recognizes a distinction
between one's racially marked experiences and one's race, the experiential
constructionist account falls short of an illuminating explanation of para-
digm cases.
The failure of the experiential account to explain ordinary intuitions
about race and passing is something a proponent of such a theory might
accept. Perhaps the ordinary conception of race is deeply flawed or inco-
herent and should be abandoned. Certainly it is sometimes necessary to
discard ordinary beliefs and concepts in light of empirical evidence or
theoretical considerations. Alternatively, the ordinary notion may need to
be abandoned because the employment of it is immoral or oppressive. The
present point is simply that a constructionist account of race that departs
significantly from the ordinary conception will also fail to explain cases of
passing that stem from this conception. Insofar as constructionists are
committed to the Passing Constraint, they are committed to going beyond
an experiential account of race.

2.2 Mills's Conceptual Constructionism

Gooding-Williams's defense of Piper falls short because an experiential
account of race does not allow for passing in paradigm cases, but the
distinction he draws between perceptual and cognitive criteria points us
towards a more adequate account. What Gooding-Williams is suggesting
is that our initial perception of someone's race may be mistaken, relative to
our background theory of racial ascription. Our view of race is, as Appiah
writes, "theoretically committed" (1996, 79). No constructionist account

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
652 NOUS

has done more to make the character of this commitment clear than Mills's
(1998) recent defense of racial constructionism. In this section, I will set out
an answer to Michaels's charge that draws upon the resources of Mills's
constructionism. I should note at the outset that the present account extends
Mills's discussion in crucial and controversial ways, and I am not at all sure
that he would endorse the specifics of the account offered here.17 I offer this
interpretation because I think it provides the best constructionist reply to
Michaels's challenge.
According to Mills, actual racial ascriptions are judgments about
whether a person meets the relevant criteria to be a member of a particular
race. These criteria are themselves the subject of an independent agreement
on the part of the community as to what are the appropriate conditions for
the application of racial labels or concepts. So Mills's racial constructionism
proceeds in two stages. At the first stage, a community agrees upon the
criteria for application of the concept or label.18 And, at the second stage,
the community determines the race of individuals by reference to those
criteria. Such an account allows us to understand how there could be
objective but constructed facts about racial membership, and how part
judgments about racial membership could be genuinely wrong (as in ca
passing). Following Mills, I will call this an objectivist account of r
To see how mere agreement on criteria can create an objective ca
consider the rules of baseball. The rules of baseball determine a batter's
strike zone, and thereby determine, for a given pitch, whether that pitc
strike. It's because we recognize the rules as determining an objective
about strikes that we can say things like 'the umpire made a bad ca
Mills's suggestion is that we can make sense of a community being mistak
about a person's race just as we make sense of a strike in baseball-b
reference to the antecedently agreed upon criteria that the community w
attempting to apply. Just what these criteria are could form the subject
social psychological investigation, but Mills suggests that bodily ap
ance, ancestry, self-awareness of ancestry, public awareness of ances
culture, experience, and self-identification are all relevant.20 Mills's d
sion makes it clear that he is interested in discovering criteria relevant t
ordinary, or folk, concept of race. I will therefore call Mills's account a f
objectivist account of race.
On a folk objectivist account, we can explain passing as long as we
make sense of mistaken judgments as to whether someone meets the
vant criteria. Here is where something like Gooding-Williams's distin
between perceptual and cognitive criteria is useful. Passing in a
category is possible because there is a distinction between those prope
that are indicative of category membership (such as easily perceptible rac
markers), and those that are central. A property is indicative of cate
membership if having the property increases the likelihood that one
member of the category. In the United States wearing a dress is indicativ

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 653

being a woman. However, it is also the case that some men wear dresses,
and many women never do. Wearing a dress is neither necessary nor
sufficient for being a woman. Similarly, having a certain color of skin or a
certain way of speaking may be indicative of one's being a member of a
particular race, even though skin color and manner of speaking are neither
necessary nor sufficient for racial membership. Other properties, however,
are more central:

A criterion p for the application of a term or concept c is central to the

extent that users of c consider p's obtaining to be a necessary or
sufficient condition for the application of c.21

For example, Appiah writes that, "There is from the very beginning unti
the present, at the heart of the system [of racial classification] a simple rule
that very few would dispute even today: where both parents are of a single
race, the child is of the same race as the parents" (1996, 77). Thus, if Appiah
is correct, ancestry is a very central criterion for folk racial ascription. A
great deal of discussion surrounds the apparently very central 'one-drop'
rule which holds that having a single black ancestor-'a single drop of black
blood'--is sufficient to make one black. Mills's various criteria range from
those that are more central to those that are less. Mills never offers an
authoritative ranking, but it is clear that he, like Appiah, holds ancestry
be very central to racial classification in the United States. He write
the U.S. racial system, at least for whites and blacks, ancestry is u
taken as both necessary and sufficient for racial membership" (1998
Mills's position is that there is a folk taxonomic system of race that
mines (more or less) who counts as being of which race. On his acc
ordinary racial concepts determine objective criteria for racial ascri
but individual judgments about the application of these concepts (and
associated labels) may be mistaken.22 Hence:

Folk Objectivist Account: A person is race R if they are the type

person that falls under the folk concept R (where this involves satisfy
ing the criteria central to the application of that concept).

The folk objectivist account allows us to understand passing. Pass

occurs when someone has the properties that are central to members
one category but has other properties that are indicative of members
another mutually exclusive category. Understanding the distinction betw
central and indicative features of a racial category allows us to see
identifying as, appearing as, and being identified as a member of race R
not sufficient to be a member of race R. Identification as a member of a
certain race, apparent race, and experiential race are merely indicative of
racial membership. For a person's race, on the folk objectivist account, is

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
654 NOUS

determined by whether or not they satisfy (at least some

central features of the concept, not by whether they are thou
the central features. Just as the umpire can make a mistake ab
pitch is a strike, so the community can make a mistake abo
particular person is a member of a race.

2.3 Thin, Not Racialist

Remember that Michaels began by suggesting that a constructionist middle
ground cannot be found and that we must think of race either as an essence
or as an illusion. By accounting for passing, Mills has shown us how to
account for the objectivity of race. But is the objectivist account now
committed to some sort of "biological essentialism"? If so, then the explan-
ation of passing has been achieved only by abandoning social construc-
tionism in favor of a biological account of race. It is easy to suspect that this
is what has occurred. For in employing the folk notion of race, the object-
ivist has imported those characteristics that the ordinary notion of race
takes to be central. Just what these are, as I said, remains open to empirical
inquiry, but it is a good bet that they will include some set of heritable
features, and in particular, one's ancestry. Moreover, the assertion that the
ordinary conception of race is committed to some sort of biological essen-
tialism is ubiquitous in theoretical discussions of race. All this suggests that
race, on the folk objectivist account, is a biological kind after all.
Concerns about essentialism are motivated by the fear that a new
account of race might implicitly raise the same troubles as now repudiated
racialist accounts did, perhaps by being inconsistent with the ontological
consensus. And it is true that a folk objectivist account would allow that
people are ascribed to races on the basis of features, at least some of which
are biological and heritable. However, this, by itself, stops far short of
suggesting that these biological and heritable traits form an interesting
kind from the point of view of biology, and a fortiori, it stops short of
racialism. Instead, a folk objectivist ought to hold that the heritable traits
that form a basis of racial ascription are part of what I called a 'thin racial
endowment'. Call any account that attributes race on the basis of such a
thin racial endowment a thin account of race.23 Because reference to thin
racial endowments does not entail racialism, thin theories avoid conflict
with the ontological consensus. And because thin theories ascribe on the
basis of traits that are independent of the choices or experiences of oneself,
or one's community, they can account for passing.
In considering the centrality of ancestry in racial ascription, Mills writes
that ancestry is "crucial not because it necessarily manifests itself in biological
racial traits but simply, tautologously, because it is taken to be crucial, because
there is an intersubjective agreement.., to classify individuals in a certain way
on the basis of known ancestry" (1998, 58). The idea is, then, that epistemically
unimportant features of persons such as ancestry or other aspects of a thin

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 655

racial endowment have been imbued with social significance by social prac-
tices. But the employment of such traits for racial ascription does not commit
the user of racial concepts to the independent explanatory relevance of the
traits. Consider another illustration. Suppose someone makes a perfect coun-
terfeit U.S. $10 bill (a molecule for molecule duplicate). This counterfeit bill
might be indistinguishable from a genuine bill, but it would nonetheless be
counterfeit because it lacks a certain kind of historical relation that a genuine
bill has (it was not printed by the U.S. Mint). Whether or not a bill has such a
historical relation is an objective fact about the bill, but the fact that we care
about such historical relations is merely a product of our monetary practice. In
fact, for other kinds of money (for example gold bricks), we may care less
about its origin and more about its composition. Neither bills of a certain sort,
nor gold, count as money except against a background practice that deter-
mines what money is. Similarly, while objective features of persons such as
ancestry may figure in practices of racial ascription, the employment of those
objective features does not entail that the features have explanatory import
independently of the ascriptive practice. So, allowing that there are objective
facts about membership in racial categories that depend in part on a person's
ancestry or heritable, biological features need not imply that such features are
explanatorily significant.
To the extent that the common sense conception of race is racialist and
takes the thin basis of racial ascription to be an indication of the truth of
racialism, to that extent a constructionist account like Mills's must abandon
it. However, there seems to be nothing to prevent the constructionist from
claiming that common sense practices of racial ascription determine racial
membership without endorsing the additional, racialist claims that may also
be part of common sense. That is, there is nothing to prevent a construc-
tionist from defending a folk objectivist theory of ascription, without
defending all the folk beliefs associated with the concept.

2.4 What is Conceptual Constructionism?

The account I have drawn from Mills is what has been called a conceptual
constructionist account.24 Just what it means to say concepts are socially
constructed can be difficult given ongoing disputes about the relationship
between concepts, their meanings, and their referents, but to explore these
issues here would take us very far a field.25 Let us say that beliefs are relations
of persons to propositions and that concepts are the constituents of these
propositions. The fact that a particular concept is used by a group of people
to understand the world is-for many and perhaps all concepts-a culturally
and locally historical fact. There is nothing inevitable about the use of concepts
like race, bamboo, river or spinal meningitis. Use of these concepts by people is
a historically and geographical local phenomenon. Among these local con-
cepts, however, we can further distinguish those that correspond to kinds of
things in the world from those that do not. The use of some concepts, while not

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
656 NOUS

inevitable, allows greater success in the explanation and pred

social, objective world. So, for example, if you are familiar
you know that an unfamiliar creature is a feline, you are like
successfully infer many things about the creature by using t
And, it seems, the best explanation of this success is that the
theories in which they figure are true, or at least approxima
This is not the case with many other concepts-concep
name in French that begins with the letter 'p' provide
inferring things about one instance from others. We should
be a range of kinds between those like feline and those like
French that begins with the letter 'p'. The ontological co
race is very far along this continuum away from the forme
latter. Hence, concepts of biological race are a poor bas
properties of members of racial groups.
Conceptual constructionists hold that

(CC1) The use of concepts of race (like that of many oth

a historically and culturally local phenomenon.
(CC2) Biological race concepts do not pick out natural, biological
kinds (i.e. the ontological consensus is true).
(CC3) Such concepts may nonetheless apply to persons who satisfy the
criteria of ascription-criteria that are independent of judg-
ments about particular cases.27

Notice that while the use of particular concepts may be culturally and
historically local, the extensions picked out by those concepts need not be.
So, for example, my concept of feline refers to local felines that I may have
interacted with, but it also refers to felines in distant times and places that I
will never interact with. Such distant felines are in the extension of feline all
the same. Folk race concepts, it would seem, conceive of race in the same
way. American folk concepts like black, white, and Asian apply-or at least
are conceived of as applying-to people in distant times and places.28 But it
is precisely because folk concepts seem to apply even to people who do not
believe they do, even among people who do not believe they do, that a
constructionist can appeal to the folk objectivist account of racial ascription
to satisfy the Passing Constraint.

3. On Not Traveling
3.1 The No Travel Constraint
Among the most characteristic and ubiquitous constructionist claims
race is the claim that 'race does not travel'. Root recently made this claim
a rather dramatic form, writing:

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 657

Race does not travel. Some men who are black in New Orleans now would have
been octoroons there some years ago or would be white in Brazil today.
Socrates had no race in ancient Athens, though he would be a white man in
Minnesota. (2000, S631-632)

Root's claim is dramatic for two reasons. First, as we just noted, commo
sense takes it that a person can fall under concepts like black, white, or Asian
even if they exist in a distant time or place. Second, common sense takes
that a person's race does not change as the person travels through space and
time. If the above discussion of the folk objectivist account is correct, this is
because folk concepts of race are ascribed to one on the basis of elements of
one's thin racial endowment that do not change from place to place or tim
to time.29 Root's claims are surprising, but such claims are quite commo
among constructionist writers though sometimes they take other forms. For
example, Mills claims that "Westerners created race in the first place, by
demarcating themselves from other 'races"' (1998, xv). Such talk implie
that race was brought into being by human decisions, and perhaps it can be
altered or destroyed by such decisions as well.
Providing an account of race that underlies such claims is important to the
constructionist because these assertions draw attention to the contingency of
the use of 'race' talk in American life and to the contingency of America
racial divisions. American society was not and is not divided along race lines
because nature made it that way, but because people acted, and continue t
act, in particular ways to maintain such a division. Constructionists emplo
such provocative claims in an attempt to show that certain facets of social lif
that might be taken for granted are, in fact, open to transformation by
collective action. But how are we to interpret such claims?
Since I have already argued that the folk objectivist account is committed
to a kind of conceptual constructionism, one possible interpretation of these
dramatic assertions by Root, Mills and others is that they mean to claim:

Concept Localism: The use of racial concepts (or some particular racial
concepts) is a culturally local phenomenon.

Thus reinterpreted, Mills's claim would be that "Westerners" created the

practice of using the race concept. And Root would be asserting that me
who are classified as black in New Orleans now would once have been
classified as octoroons. The concept localism reinterpretation of these c
has at least two enormous advantages. To begin with, constructio
explains passing precisely by appeal to the objectivity of race as determ
by the agreed upon criteria for the use of race concepts. So, the co
localism reinterpretation allows us to make sense of these addition
structionist claims by Mills, Root, and others by reference to th
conceptual constructionist account I've already attributed to Mills.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
658 NOUS

concept localism is true. It is the case that the use of rac

concepts varies widely from place to place and time to tim
features make the concept localism interpretation of Root, M
attractive indeed.
Despite these advantages, I think the concept localism reinterpretation fails.
It amounts to a reinterpretation in that it attributes to constructionists
widespread use-mention error. To interpret them as merely asserting concept
localism when they say things like 'race was created' or 'race does not travel' i
to suggest that they are not talking about race at all but the concept of race.
And this latter claim is far less interesting and far less counterintuitive than th
former. Moreover, some claims constructionist thinkers make block this rein
terpretation. For example, Mills writes, "Race may not be real in the sense that
racial realists think or even like, but it is real enough in other senses" (1998,
66). This passage makes little sense unless it is interpreted as being about race
and not the race concept. So even if many constructionist claims are confu-
sions of use and mention, a variety of core examples are not. I find thes
reasons compelling, and so I think that in making claims about race no
traveling, being created, transformed or destroyed, constructionists are com-
mitted to something more than just concept localism. They are committed to:

The No Travel Constraint: On a constructionist account of race, a

person's race is dependent upon the culture in which they live.

For the remainder of the paper, I will assume that claims about race
traveling or race being created, transformed, or destroyed reflect an
allegiance to this stronger constraint.

3.2 Root's Principle

But how is such a constraint to be satisfied? Fortunately, Root is quite clear
both about his commitment to the No Travel Constraint and about how he
thinks it can be satisfied. Continuing his previous passage, Root writes:

Where R is a race, a person is R at a site only if R is used there to divide people.

Because the ancient Greeks did not divide people by race, there were no races in
Athens then and, of course, no differences between people in morbidity or
mortality based on race there.

Biological categories are different. Whether we are Homo sapiens, male, or

Rh positive does not depend on how we categorize ourselves or what we
understand of biology. With sex and blood type, how people are classified,
the category they belong to, does not depend on their self conceptions or on
whether they recognize the category, while with social classification self con-
ceptions are central. A group of people must divide themselves by race but not
blood type in order to have one.30

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 659

The asymmetry between genuine biological categories and race stems from
the role that the classification of race plays in constituting or creating race.
Root's claim that "where R is a race, a person is R at a site only if R is used
there to divide people" needs to be clarified. Note that the same term 'R'
appears on both sides of the conditional, but it is not clear what it would
mean to use a race at a site to divide people. What Root intends is that,

Root's Principle: Where R is a race, it is a necessary condition on a

person being R at a site that the concept of R is used at that site to
divide people.

When combined with concept localism, Root's Principle entails that race is
culturally local, and thus it offers an explanation of how constructionists
can hope to satisfy the No Travel Constraint.

3.3 Interactive Kinds

Questions remain about how we should understand Root's principle. On
way to understand it is provided by recent work by Ian Hacking (1995a
1995b, 1999) on what he calls interactive kinds. Interactive kinds are kinds of
persons that come to be instantiated when people causally interact with th
regimes of labeling-practices of labeling and differentiating people-to
which they are subjected. The principle idea of the interactive kind account
is that practices of systematically labeling persons in a culture have causa
effects upon those so labeled.31 Thus, we get:

The Interactive Kind Principle: In order for a person to be race R, they

must be at a site where the concept of R is used to divide people since
being labeled by a term expressing the concept R is causally necessary
to becoming an R.

To be 'labeled' by a term expressing a concept R involves more than fallin

in the extension of R. One must see oneself, or be seen by others, as falling
under the concept in order for a site's practice of using concept R to hav
its causal effects on an individual. We have already seen that Piper defend
an experiential account of race that endorses the Interactive Kind Princi
ple. Recall that an experiential account of race holds that a person is a
member of race R if they have R-typical experiences as the result of having
been labeled and differentiated by a community.32 Since one cannot hav
the appropriate experiences except in a community that divides itself i
such a way as to create them, the experiential account shows a way that the
Interactive Kind Principle could be true.
Similarly, a defender of an identification account might endorse thi
principle. Remember such an account holds that a person is race R if the

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
660 NOUS

identify as a member of race R where that involves 'believ

acting' in R appropriate ways. Having R-typical beliefs a
R-typical actions would, on such an identification accoun
internalizing locally held beliefs about what sorts of per
Hence one could not be an R except at a site where the c
used to divide people.
This interpretation of Root's Principle allows us to see the w
racial theorist could satisfy the No Travel Constraint. The
simple: consider certain causal effects of practices of racial la
features of race. Then, since concept use is culturally local, rac
Accounts respecting the Interactive Kind Principle (interactiv
also allow us to make sense of claims such as that race was cre
could be destroyed or transformed. What such claims require
munity creates, destroys, or transforms a labeling practice or i
Unfortunately, while interactive accounts can satisfy
Constraint, we have already seen (section 2.1) that they do
Passing Constraint. They fail to satisfy the Passing Cons
paradigm cases of passing seem to indicate a person could
race R even if they do not conceive of themselves as being a m
R and even if they are not treated by others as being a memb
short, passing may involve someone's being an R but esc
effects of being labeled as an R. If interactive kind accoun
way to make sense of the No Travel Constraint, then no accou
both the Passing Constraint and the No Travel constraint.

3.4 Institutional Kinds

While interactive kind accounts cannot satisfy the Passing Constraint, t
accounts of race like the folk objectivist account do not satisfy the N
Travel Constraint. This is not surprising, as the two accounts of race a
responses to two different constraints, one suggested by the folk experienc
of passing and the other by the needs of racial theorists to draw attention
the contingency of racialized living. The question is, can any account of rac
satisfy both constraints?
The answer seems to be yes. Interactive kind accounts read the wor
'necessary' in Root's Principle as implying the causal necessity of the us
a certain label. But the principle can also be read as expressing log
necessity. Thus, we get:

The Institutional Principle: Where R is a race, it is a logically necessary

condition on a person being R at a site that the concept of R is used a
that site to divide people.

Any account of race that respects the Institutional Principle satisfies the N
Travel Constraint since race exists only at sites where the concept R is used

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 661

to divide people, and concept localism is true. We can simply add the
Institutional Principle to the folk objectivist account of race to get:

Folk Objectivist Institutional Account: A person is race R iff

(1) He or she is the type of person that satisfies the criteria central to the
application of a folk racial concept.
And (2) the person is at a site where the concept R is used to divide people.

This account satisfies the Passing Constraint in the same way the folk
objectivist account does: by giving up the requirement that a person actually
be labeled as an R, and embracing only the requirement that they be the
type of person that falls under the concept R. And the account satisfies the
No Travel Constraint by limiting the application of a race concept R to sites
where the concept is used to divide people.
At first look, adding the Institutional Principle to the folk objectivist
account may seem quite arbitrary. After all, if the folk concept of race has
persons in its extension across time and space, it seems ad hoc to simply
impose a restriction that limits the extension of racial terms to sites where
the concept of R is used. However, such restrictions seem less out of place if
we think of race as a social institution. On many accounts, social institu-
tions-things like being a licensed bass fisherman, a wedding ceremony, or
money-essentially involve people who think about things under the con-
cepts licensed bass fisherman, cocktail party, and money. Here's John Searle's
way of putting it:

... For social facts, the attitude that we take toward the phenomenon is partly
constitutive of the phenomenon .... Part of being a cocktail party is being
thought to be a cocktail party; part of being a war is being thought to be a
war. This is a remarkable feature of social facts; it has no analogue among
physical facts. (1995, 33-34)

According to Searle, institutional facts of this sort are created only when a new
status is imposed on an object as the result of collective concept use. Applying
this kind of analysis to the case of race, the Institutional Principle seems less
arbitrary. The proposal is that we understand race as a kind of institutional
fact created by collective understandings of who counts as what race here. By
doing this, we arrive at a constructionist account that can apparently satisfy
both the Passing Constraint and the No Travel Constraint.

4. The Reality of Race

The folk objectivist institutional account seems to be just what is required to

satisfy both the Passing Constraint and the No Travel Constraint. But

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
662 NOUS

accounts of race that respect the Institutional Princ

accounts) fail to satisfy a third constructionist commi
constraint is suggested in claims like Mills's assertions t
"exists (and moves people)." How are we to interpret su
nature of reality is a much discussed philosophical topic
little to add here. But if we look at constructionist claims about the char-
acter of reality, we see that in discussing the reality of race, they ha
something very particular in mind. Mills argues that one's race in contem
porary America has a real and important impact on a variety of aspects
one's life. And Root writes,

Race affects income, housing, and healthcare, and these, in turn, affect
health.... As a result race can enter into many statistically robust biomedical
generalizations even though there are no biological races. (S629)

Though they express it in different ways, both Mills and Root believe ra
enters into a variety of nonaccidental generalizations, and that one's race in
contemporary America differentially affects one's life chances in ways that
are not explained by one's biological make up.33 Root, in particular, is
concerned to show that a constructionist account of race can figure in
generalizations in social scientific enterprises, and he employs the term
'real' of a social category just in case such a category has a causal influen
on members' lives.34 Recall also that, at the outset, we noted the conserv
tionist constructionists' insistence that it is the reality of race that mak
metaphysical doctrines that eschew reference to racial properties inad
equate. This all suggests that Mills, Root, and other constructionist
endorse the following:

The Reality Constraint: On a constructionist theory of race, race should

(a) Figure in explanations of the differential impacts on individuals in

racially divided settings.
And (b) Figure in the formulation and explanation of true, nonaccidental
generalizations.35

In the remainder of this section, I argue that accounts of race that respe
the Institutional Principle (institutional accounts) fall afoul of the Realit
Constraint in two ways. First, institutional accounts of race are empty
that they apply to no one. As such, institutional race does not explain an
facets of racialized living in contemporary America. Second, racial genera
izations require the use of thin concepts that travel. I argue first that
while institutional race may figure in some racial generalizations, the
needs of confirming such generalizations compromise the satisfaction o
the No Travel Constraint by the use of concepts that pick out (apparentl
racial) groups across times and places. Moreover, some ordinary racial

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 663

generalizations seem to require that race travels. Because institutional

race does not travel, reference to it cannot figure in such generalizations.
Institutional accounts of race thus fail to satisfy the Reality Constraint.

4.1 The Emptiness of Institutional Race

Institutional accounts of race differ in a key way from the ordinary concepts
of the folk objectivist account, for an institutional account of race incorp-
orates an additional necessary condition: the institutional principle. Thus, a
folk objectivist institutional account of race is not intended as an elucidation
of ordinary racial concepts. Rather, the account stipulates a new race
concept in order to satisfy a theoretical need (the No Travel Constraint).
For this very reason, however, institutional racial concepts do not apply to
actual people (save perhaps among those who know some academic phil-
osophers). The argument for this is simple:

El. The institutional principle requires that for a person to be a

member of race R, they must be at a site where the concept of R
is employed.
E2. But, where the concept of R is a racial concept constituted, in part,
by the institutional principle, there are no sites (except possibly
near philosophers) where such concepts are used.
E3. Thus, no one (or almost no one) is R.

Because they are empty, institutional accounts of race have little use in
discussions of actual processes of racial differentiation. They cannot play
a role in explaining the apparent reality of race in everyday life, because no
one in everyday life uses such concepts. And while perhaps there are true,
nonaccidental generalizations involving uninstantiated properties, it is hard
to see the importance of formulating such generalizations or making refer-
ence to such properties.

4.2 Racial Generalizations. The Need for Race that Travels

Important claims regarding race in social science and social, political, and
moral theory require a concept that travels. These causal and normative
claims appear to require reference to other persons that fall under racial
concepts but who live in a culture in which those classifications are not used
(and thus do not fall under non-traveling race concepts). If this is correct, it
weighs not only against institutional accounts of race, but against any
account of race that satisfies the No Travel Constraint (e.g. interactive kind
accounts). While we can frame some racial generalizations with institutional
race concepts, the defender of institutional race continues to need other
concepts-apparently thin racial concepts--that do travel.
Consider Root's claim that, "Blacks are seven times more likely to die of
tuberculosis in the U.S. but not in Great Britain."36 Root argues that higher

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
664 NOUS

tuberculosis rates among American blacks are not the resul

biological endowments but rather are the product of rac
States. Thus he endorses the causal claim:

Cl. American racism causes American black persons to die of tub

culosis at much higher rates than American white persons.

This is an important claim that substantiates the view that Americ

society is in certain respects white supremacist. But to legitimize this ca
claim, the social scientist needs to be able to identify control grou
blacks that live in other cultural contexts where racism is not as severe and
see if their tuberculosis levels are lower. Why would this be? Suppose, fo
example, a survey of life expectancy data showed that U.S. Presidents hav
shorter average life spans than other U.S. citizens. We might then endors
the claim,

C2. The stress of the U.S. Presidency has a destructive effect on the
health of those who hold that office.

In this case, our control group would be made up of ordinary persons who
had not occupied the social role of U.S. President. We do not need to
assume that the property of being a U.S. President obtains of persons
among whom the concepts are not used, so why would we have to make
such an assumption in the case of a racial concept?
The reason is that the legitimation of causal hypotheses requires the
exclusion of alternate possible causes. This is the point of Root's compari-
son to blacks in Great Britain (as opposed to, for example, whites in Great
Britain or just any group of people anywhere). His argument shows the
higher tuberculosis rates among blacks in the United States are social rather
than biological in origin which substantiates his hypothesis that racism is the
cause. The comparison provides important evidence to show that the reflex
racialism of everyday thinking is mistaken, and that the results of such
racialism can be tragic. But the comparison only works if we pick out the
same group picked out by the American racial concept. Otherwise, the
comparison would not have eliminated the biological explanation.37'38 Because
one aim of social scientific research is undermining racialism and revealing
the socially contingent character of racial difference, institutional accounts
of race will not do all the work required here. When we state generalizations
in terms of institutional race, we continue to need a way to identify members
of racial categories in other contexts free of the consequences of racial
categorization. This need, by itself, does not violate the Reality Constraint,
as the defender of institutional race can simply insist that the groups picked
out cross-culturally are not properly considered racial groups. However, the
need for cross-cultural identification of groups does reveal a tension in

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 665

attempting to satisfy the Reality Constraint and the No Travel Constraint

simultaneously. It reveals that even on an institutional account, satisfying
the Reality Constraint requires recognizing at least that some sort of race-
like group travels.
But some ordinary generalizations seem to require race itself to travel.
For example, a variety of ordinary normative claims imply generalizations
requiring racial properties that travel. Consider the claim that, "Asians,
blacks, Hispanics, and whites would all be better off in a society free of
racial categorization." If the presence of American practices of racial divi-
sion is a logically necessary condition for the application of American racial
concepts, then these alternatives are impossible. For example, on an institu-
tional account of race, Asians would not be better off in a society free of
racial categorization since Asians would not even exist in such a society.
One might object that such normative claims can eschew use of concepts
making reference to racial properties that apply across cultures in favor of
concepts that apply only to actual people in, for example, the contemporary
United States.39 The idea is that when we say "Blacks, whites and Asians
would all be better off in a society free of racial categorization," we construe
it to mean something like, "the actual people who are labeled by terms (or
the concepts associated with) 'black', 'white' and 'Asian' would be better off
in a society free of racial categorization." We simply imagine a case in which
the actual people who live in the contemporary, racialized United States
exist in a new setting-one free of racial labeling. Again, the hope is to avoid
implying there is some extra-cultural property of being black, white, or
Asian that members of racial groups share.
The problem with this strategy is that the sort of normative claims in
question may extend beyond the scope of actual people. For example,
consider the claim that, "future generations of blacks, whites and Asians
would be better off in a society free from racial classification." This latter
claim extends to nonexistent members of American racial categories. The
claim makes sense because folk racial concepts carry with them certain
criteria that determine what (actual and possible) people fall under them.
These cases seem to require that we employ, at least for the purposes of
making such statements, a concept picking out a group that does travel.
What is this account? The most obvious option is to choose a thin account
of race, for example, the folk objectivist account. Thus, we could imagine
that claims about the effects of racism or racial labeling on American
racial groups could be corroborated by looking at members of those same
racial groups in times and places free of racialized treatment. But other
social scientific studies might choose a different thin concept, something like
what we called apparent race. This latter notion remains a viable candidate
for social science because it may be, as Root suggests, that "difference in
health or socioeconomic standing correlate with perceived race" rather than
race as construed by the folk objectivist account (2000, S635).

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
666 NO US

Whatever thin account is chosen, consideration of the abo

suggests that reference to persons in other times and places
labeled by ordinary American racial terms is a key element o
about race in America for social scientists and social theorist
state generalizations in terms of institutional race, we still retai
(and conceal the use of) a thin, traveling concept. And inst
concepts make generalizations about non-actual members of
impossible, thus running afoul of the Reality Constraint.

5. Beyond a Univocal, Constructionist Account of Ra

If the preceding arguments are correct, there is no univocal acco

that can do everything constructionists want it to do. This c
reminiscent of Appiah's claim that "there is nothing in the w
do all we ask 'race' to do for us" (1995, 75) that we might w
should, like Michaels and Appiah, be led away from constru
towards skepticism. However, this is not the only option. Fa
difficulty of simultaneously satisfying the adequacy const
account of race, we can also choose to give up one or anoth
constraints. In fact, each of the metaphysical accounts consi
viewed as having abandoned a constraint:

1. Thin Race (e.g. a folk objectivist account).41

(a) Abandon the No Travel Constraint.
(b) Employ a thin account of race both to account for Pa
for the Reality of Race.

2. Interactive Race (e.g. an experiential account).

(a) Abandon the Passing Constraint.42
(b) Employ a causal effect or combination of causal effect
concept use to satisfy the No Travel Constraint.

3. Institutional Race (e.g. a folk objectivist institutional acc

(a) Abandon the Reality Constraint.
(b) Employ a folk objectivist institutional account of race
the Passing and No Travel Constraints.

4. Skepticism: Abandon all the constraints because race does

This list suggests four ways a race theorist might proceed in th

three mutually inconsistent constraints, and no doubt there are
can we choose among these accounts? Which constraint shall
In proposing to abandon one or another constraint, a race
insist that the motivations behind a constraint can be satis
other account that is nonetheless not an account of race. F

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 667

Appiah's (1996) racial skepticism is accompanied by a theory of racial

identity intended to account for identification with racial concepts and the
causal effects of such identification. He thus tries to satisfy someone con-
cerned with the real impacts of racial classification, but without agreeing
that race exists. Similarly, defenders of a thin account of race may insist that
although race concepts can be cross culturally applied (i.e. race travels), the
causal effects of the practice of employing such concepts and their asso-
ciated labels are culturally local. Such a theorist would thus try to satisfy the
theoretical need driving the No Travel constraint, but without allowing that
it is race that fails to travel. And defenders of interactive or institutional
accounts may suggest that the motivations behind the Passing Constrain
and the Reality Constraint can be answered by employing concepts ascribe
according to thin racial endowments, but that such concepts ought not to be
considered race concepts.
But for present purposes, we need not decide at all. While the four
positions appear to represent different approaches to the metaphysics of
race, it is hard to find a substantial point of metaphysical disagreemen
among them. In particular, all endorse the following general theses:

1. The employment of particular racial labels and their associated

concepts (e.g. 'white', 'black', 'Hispanic', 'Asian', etc.) is a culturally
local practice.
2. These concepts impose particular, objective standards for racial
classification.
3. The causal effects of the employment of these concepts-including
effects on experiences, identifications, and a variety of relational
properties-are also culturally and historically local.
4. These causal effects can profoundly affect the life chances of those
who fall under the concepts.

While some disagreements remain about the details of these four theses (for
example, on the exact nature and extent of the causal effects of racial
classification), it is safe to say that these four features, combined with th
denial of racialism form an expanded ontological consensus. Indeed, if our
discussion above is correct, then to deny that any of these four features
obtains is to make some sort of factual mistake. The four metaphysical
accounts above are each grounded in a set of intuitions (perhaps semantic
and normative intuitions) about which features recognized in the expande
consensus should figure in fixing the meaning or referent of genuine racial
terms and concepts. Fortunately, it is difficult to see anything metaphys
ically important that hinges upon the disagreement. Even if one of the four
accounts above correctly captures the best or the authentic concept of race,
there is no reason why theorists should not introduce new, technical notions
to suit various theoretical needs. Even if these technical notions do not

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
668 NOUS

capture the meaning of folk race concepts or 'race' talk,

focus our attention on dimensions of the folk practice o
causal effects.43
In light of the failure to find a univocal constructionist a
we must abandon the attempt to provide an account of r
satisfies all the constraints suggested above. Since all pa
expanded ontological consensus, and all parties should agree
ations behind the various constraints are important, th
which motivations should be satisfied with a genuinely
which should be satisfied by some or another nonracial a
etical decision about the correct meaning of racial terms, bu
significance for the metaphysics of race. Instead of focusin
account of what race is or what racial concepts mean, we
kinds of conceptual apparatus do we need to discuss racia
racially associated phenomena in historical and contempo
exchange the question of whether and how race exists f
developing an adequate metaphysical theory distinguishing as many
accounts of race or racial phenomena as are needed to serve all our func-
tional needs-including the various dimensions of racial identification,
experience, appearance, and folk classification-so that their practical,
social, and ethical significance can be discussed. Only in such a project of
theoretical refinement are we likely to shed the persistent mistakes of
ordinary racial thinking while continuing to refer to the world in ways
that satisfy a multiplicity of theoretical needs.44

Notes

1On some of the accounts of race discussed below, they are also disputes over the value o
race itself.
2Other, moral, arguments for eliminativism may be premised on the truth of racia
skepticism. For example, because racial properties do not exist, according to some skeptic
eliminativists, it is morally arbitrary to use racial labels to differentiate people.
3 Important discussions by social theorists preceded these accounts, including, e.g. Stuart
Hall 1987a, 1987b; Omi and Winant 1986, 1994.
4 1996, 54. To make Appiah's characterization more precise we need to distinguish between
heritable differences that are mediated by the environment, and those that result more directl
from a genotype. Heritability measures the ratio of variance in a characteristic in an environment
that is due to genes to the total variance in the environment. If a community systematically
tattoos the hand of every person with a cleft in their chin, then the heritability of the tattoo wou
be high. Nonetheless, the presence of the tattoo on the hands is the result of a social policy.
Racialism assumes that genetic or biological makeup is the mechanism of heritability.
5 Robin Andreasen (1998, 2000) and Philip Kitcher (1999) have recently offered relational
or historical accounts of racial categories as biological kinds, with Andreasen maintaining tha
races are clades of a phylogenetic tree and Kitcher arguing that races are reproductively isolate
populations. While I will not discuss these accounts here, it is important to note that while o
the racialist account, races are differentiated by intrinsic, heritable and explanatory features

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 669

(e.g. 'blood' or genetic material), Andreasen and Kitcher identify them historically or relation-
ally. And while Andreasen and Kitcher argue that race can be understood as a legitimate
biological category, they nonetheless deny racialism. For present purposes, I will count the
relational and historical features discussed by Andreasen and Kitcher as part of one's thin racial
endowment. Note that because these accounts are naturalist but not racialist, they call into
question the need for a constructionist 'third way' between naturalism and skepticism. I will
remain focused on constructionism, but I think the distinction between naturalistic and thin
constructionist approaches to race is not as significant as it has been made out to be. (See also
fn. 23 below and Section 5.)
6 Cf. Lewontin 1972, 1982; Cavalli-Sforza, Menozzi, and Piazza 1994 for discussions of
human diversity.
7While there is a consensus, there are still dissenters. E.g. Albert Mosley (1999) is skeptical
of Appiah's claim that race does not exist. But, it is unclear whether Mosley's arguments would
lead him to endorse racialism as I have characterized it here, or some version of what I call a
thin account of race below.
8 Issues get quite complex here because 'racial superiority' may be relativized to a domain.
For example, it might be part of a traditional racist doctrine that blacks are intellectually or
morally inferior but in some other way (e.g. physically, musically, etc.) superior to whites.
Because some putatively anti-racist but racialist accounts of racial difference (e.g. some of those
of the Negritude movement) also endorse the notion of differential racial gifts, it may be
difficult to decide whether or not such doctrines are racist.
9 Many philosophers and others inspired by a Kantian moral conception may wish to remain
neutral on the biological issue, and instead insist that even thick, biological racial endowments
would be morally and politically irrelevant. I am sympathetic to this claim, but I don't intend to
take up this question here. I do believe that if thick, biological racial endowments did exist they
would pose serious practical challenges for achieving political justice even if the best moral theory
says they ought not to be relevant. I am grateful to Jesse Taylor for discussion on this point.
0o Notice also that depending on how we answer these questions, some of the accounts that
follow will-or will not--qualify as accounts of race. In what follows, I will ignore this issue,
and use the term 'race' somewhat promiscuously for a variety of accounts that have been
proposed in connection with race.
11E.g. Piper 1992, Mills 1998, Root 2000.
121994, 758
13 One might think this is the position that Appiah (1996) defends. While Appiah defends
an identification account of racial identity, he distinguishes racial identity from race. Moreover,
Appiah agrees with Michaels that identification is not sufficient for falling under racial labels,
writing that "It is because ascription of racial identities.., is based on more than intentional
identification that there can be a gap between what a person ascriptively is and the racial
identity he performs" (1996, 79). Appiah's account of racial identity thus holds that identifica-
tion has certain preconditions that are outside the scope of one's choices. Applied to race,
Appiah's account of racial identity presupposes a practice of ascription whereby persons are
assigned to races-perhaps an account like the one I will attribute to Mills below. And it is this
prior practice of racial ascription that Appiah employs to make sense of passing. Similar things
can be said of Iris Marion Young's identification account of racial social groups (1989).
Gooding-Williams's account of 'black persons' is quite explicit about this presupposition
(1998, 22ff), as is Outlaw's (1995, 1996) account of race.
14Michaels 767; Piper 30-31. Piper does not say that this experiential criterion is 'what
makes blacks black' as Michaels writes. Rather, Piper writes that the shared experience of being
visually or cognitively identified as black "joins me to other blacks,... and other blacks to
another [sic]" (30). Thus, it could be that Piper does not intend this as a criterion of blackness.
Nonetheless, I join Michaels and Gooding-Williams in assuming she does.
'15Twain 1976.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
670 NOUS

16The Phipps case is discussed in Michaels (1994, 764). Williams 1995.

cites Twain and Williams for similar reasons.
17 Below I argue that Mills is committed to incompatible constraints on a constructionist
theory. Thus the account I develop in this section is, at best, only one consistent strand of Mills's
theory.
8 The first stage may range over a historical period, such that the agreement is reached via a
distributed process in which people learn to use racial concepts and labels in particular ways.
19 Sometimes the 'objectivist' label is used to indicate naturalist or realist accounts of race
(e.g. Harris 1999b). The present use of 'objectivist' holds racial ascriptions to be objective in
J.L. Mackie's (1977) sense: "Given any sufficiently determinate standards, it will be an objective
issue, a matter of truth and falsehood, how well any particular specimen matches up to those
standards" (26). Such truths are, in John Searle's terminology, ontologically subjective but
epistemically objective (1995, 8).
201998, 50ff.
21 Note that it is possible for p to be a compound or disjunctive property. Thus, to say that
some properties of race are more central than others is not to say that races have simple
necessary and sufficient conditions.
22 A number of theorists have suggested that folk notions of race cannot play the role that
Mills assigns them. Such theorists assert that racial ascriptions are actually incoherent (e.g.
Piper 1992, p. 31), or alternatively, are not rule governed practices but rather arbitrary exercises
of power (e.g. Frye 1983). While there is no space to discuss these objections here, if they turn
out to be correct, they problematize an account like Mills' that relies on the coherence and
objectivity of folk practices of racial ascription.
23Above (fn 5), I mentioned proposals by Robin Andreasen (1998, 2000) and Philip
Kitcher (2000) suggesting that race is a biological kind. Like Mills, Andreasen and Kitcher
avoid a commitment to racialism by employing only thin features in the assignment of racial
membership. So all three are thin theories. This similarity suggests more common ground between
a constructionist theory like Mills's and thin naturalist accounts than initially meets the eye.
24This sort of position has been distinguished previously by Griffiths 1997, Andreasen
1998, and Hacking 1999.
25 Throughout the discussion, I assume with others that concepts are individuated by some
subset of the beliefs in which they figure, particularly beliefs governing the conditions of
application for the concept. Some dispute that concepts can be so individuated (e.g. Fodor
1998). I take it that nothing important hangs on the outcome of this dispute for the relevant
claims could be restated as being about subsets of beliefs rather than concepts.
26This line of argumentation has been prominent among scientific realists. E.g. "it is
impossible to explain scientifically the instrumental success of scientific methods except on
the hypothesis that in mature sciences the background theories that determine methods are
approximately true of unobservable (as well as of observable) entities .... " (Boyd 1991, 14). Cf.
Mallon (2003) for an application of this sort of reasoning to social kinds.
27To these, Mills's conceptual constructionism adds a fourth feature, namely that the
criteria for the application of race concepts (e.g. the 'one drop' rule) are selected, at least in
part, according to their ability to serve the interests of those in power. For present purposes I
will focus on the weaker construal of constructionism.
28 This is not to say that they apply to all people in all times or places.
29In describing thin racial endowments, I included "various relational properties including
one's ancestry which are extrinsic to the individual, but are heritable." I am presuming that
these relations remain constant for individuals across times and places as I assess Root's
account. If we vary these relations when 'traveling' from one cultural-historical setting to
another, then race will not travel. I ignore this possibility here because it seems clear that it is
not this way of failing to travel that Root and other constructionists have in mind.
30 Root 2000, S632.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 671

31 For a fuller discussion of the causal mechanisms involved, see Mallon (2003) Section 3.
32 Piper's experiential account is explicitly an account of what is shared by blacks. Note
that I have generalized it to an account of any race R.
33 By 'nonaccidental generalizations', I mean generalizations that are not mere descriptions
or coincidences, but reflect some sort of underlying causal regularity.
34 Thus, the 'reality' of a category, in Root's usage, is distinct from its existence. He writes,
"Should we divide but not regulate by race, we would retain the races but not conserve their
reality" (S635).
35 To say that some racial generalizations are true is not to say that they obtain of each
person categorized under a racial concept.
36 Root 2000, S633.
37It might be thought that the biological alternative does not need to be eliminated
because, as we know from the ontological consensus, the biological alternative is false. But
biological thinking is much more resourceful than this. Because what I have called 'thin' racial
endowments are widely recognized, the temptation to expand the content of such endowments
to include additional factors needs to be constantly checked lest it give rise to 'biological'
explanations for the effects of social oppression. The only way to discredit such explanations is
to subject them to persistent empirical scrutiny.
8 Notice that a parallel argument suggests that in the case of U.S. Presidents our control
group should be made up of men. Because sex is known to effect life expectancy, a defender of C2
would want to exclude the possibility that the shorter average life expectancy of U.S. Presidents
results from the sex of the actual role occupants instead of the stress attached to the office.
39 Michael Root and Paul Taylor have independently suggested this approach to me.
40 Note that it is compatible with this claim that we avoid use of racial terms. The key point
is that we need to employ some or another terminology to make reference to actual and possible
persons in other times and places that fall under racial concepts.
41 This group also includes thin naturalist accounts that assign persons to racial groups
according to their thin racial endowments. Whether such an account is a kind of construc-
tionist, or an alternative to it, is a question I do not address here.
42 Alternatively, an interactive account may allow a limited form of passing while aban-
doning the attempt to handle the cases I discuss in Section 2.1. Such a theorist might insist that
they do not abandon the Passing Constraint, only the project of accounting for all the putative
actual cases of passing by using a racial concept.
43 For example, Sally Haslanger (2000) has recently offered a relational account of gender
and race in an attempt to focus theoretical and practical attention on certain morally relevant
features of gender and racial classification. And while Haslanger intends that her account of
race be accepted as an important (and perhaps even the most important) account of what we
should care about when we talk about race, it is not her claim that her account captures what
we ordinarily mean by 'race' talk.
44I am very grateful to the following for helpful discussions regarding earlier versions of
this work: Lori Alward, Robin Andreasen, Max Deutsch, Robert Gooding-Williams, Joe Lau,
Aaron Meskin, Elijah Millgram, Ram Neta, Michael Root, Jesse Taylor, Paul Taylor, and two
anonymous referees for this journal.

References

Andreasen, R. 0. 1998. A New Perspective on the Race Debate. British Journal of t

Philosophy of Science 49:199-225.
2000. Race: Biological Reality of Social Construct? Philosophy of Social Science 67
(Proceedings): S653-S666.
Appiah, K. A. 1995. The Uncompleted Argument: Du Bois and the Illusion of Race.
L. A. Bell and D. Blumenfeld (1995). Pp. 59-78.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
672 NOUS

. 1996. Race, Culture, Identity: Misunderstood Connections. In Col

Political Morality of Race, Ed. K. A. Appiah, A. Guttmann, pp. 192
Princeton University Press.
Baker, H. A. Jr., M. Diawara, and R. H. Lindeborg. Eds. 1996. Black Britis
Chicago: University of Chicago Press.
Bell, L. A. and D. Blumenfeld. Eds. 1995. Overcoming Racism and Sex
Rowman and Littlefield.
Boyd, Richard. 1991. Confirmation, Semantics, and the Interpretation of Scientific Theories. I
The Philosophy of Science. Ed. R. Boyd, Philip Gasper, and J. D. Trout. Cambridge
MIT Press.

Cavalli-Sforza, L .L., P. Menozzi, and A. Piazza. 1994. The History and Geography of
Genes. Princeton, NJ: Princeton University Press.
D'Souza, Dinesh. 1996. "The One-Drop-of-Blood Rule." Forbes. Vol. 158 No. 13. P. 4
Fodor, Jerry. 1998. Concepts: Where Cognitive Science Went Wrong. New York: Cla
Press.

Frye, Marilyn. 1983. On Being White: Toward a Feminist Understanding of Race and Race
Supremacy. The Politics of Reality: Essays in Feminist Theory. Freedom, California:
Crossing Press. Pp. 110-128.
Goldberg, D. T. 1990. Anatomy of Racism. Minneapolis: University of Minnesota Press.
Gooding-Williams, Robert. 1988. Race, Multiculturalism and Democracy. Constellations 5:
18-41.

Griffiths, Paul E. 1997. What Emotions Really Are. Chicago: University of Chicago Press.
Hacking, Ian. 1995a. The looping effects of human kinds. In Sperber (1995). Pp. 351-394.
1995b. Rewriting the Soul: Multiple Personality and the Sciences of Memory Princeton,
NJ: Princeton University Press.
1999. The Social Construction of What? Cambridge, USA: Harvard University Press.
Hall, Stuart. 1987a. [1996]. Minimal Selves. Reprinted in Ed. Baker et al. (1996). Pp. 114-19.
. 1987b. [1996]. New Ethnicities. Reprinted in Ed. Baker et al. (1996). Pp. 163-172.
Harris, Leonard. Ed. 1999a. Racism. New York: Humanity Books.
. 1999b. "Introduction." In Harris (1999a).
Haslanger, Sally. 2000. Gender and Race: (What) Are They? (What) Do We Want Them To
Be? Nous 34:1:31-55.

Kitcher, Philip. 1999. Race, Ethnicity, Biology, Culture. In Harris (1999a). Pp. 87-12
Lewontin, R. 1972. The apportionment of human diversity. In T. Dobzhansky, M .K
W. C. Steer (Eds.) Evolutionary Biology, vol. 6. New York: Appleton Century
Pp. 381-198.
1982. Human Diversity. New York: Scientific American Press.
Mackie, J. L. 1977. Ethics: Inventing Right and Wrong. London: Penguin Books.
Mallon, R. 2003. Social Construction, Social Roles, and Stability. In Ed. F. Schmitt
Pp. 327-353.
Michaels, Walter Benn. 1992. Race into Culture: A Critical Genealogy of Cultural
Critical Inquiry 18: 655-685.
1994. The No-Drop Rule. Critical Inquiry 20: 758-769.
Mills, C. 1998. Blackness Visible: Essays on Philosophy and Race Ithaca: Cornell Un
Press.

Mosley, Albert G. 1999. Negritude, Nationalism, and Nativism: Racists or Racialists? In Harris
(1999a). Pp. 74-86.
Muir, D. 1993. Race: The Mythic Root of Racism. Sociological Inquiry 63: 339-350.
Omi, M. and Winant, H. 1994. Racial Formation in the United States: From the 1960s to the
1990s. New York: Routledge.
1986. Racial Formation in the United States: From the 1960s to the 1980s. New York:
Routledge and Kegan Paul.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms
 Social Constructionism and the Metaphysics of Race 673

Outlaw, L. 1990. Toward a Critical Theory of "Race". In D. T. Goldberg (1990). Pp. 58-82.
.-. 1995. On W.E.B. Du Bois's "The Conservation of Races". In L. A. Bell, D. Blumenfeld
(1995). Pp. 79-102.
1996. On Race and Philosophy. New York: Routledge.
Piper, Adrian. 1992. Passing for White, Passing for Black. Transition 58: 4-32.
Root, Michael. 2000. How We Divide the World. Philosophy of Social Science 67 (Proceedings).
628-639.

Searle, John. 1995. The Construction of Social Reality. New York: Free Press.
Schmitt, F. Ed. 2003. Socializing Metaphysics. MD: Rowman and Littlefield.
Sperber, Dan. Ed. 1995. Causal Cognition. New York: Clarendon Press.
Sundstrom, Ronald. 2002. Racial Nominalism. Journal of Social Philosophy Vol. 33. No.
pp. 193-210.
Twain, Mark. 1976. Puddn'head Wilson, Vol. 1. Ed. Langston Hughes. New York: Bant
Doubleday Dell Publishing Group.
Webster, Yehudi. 1992. The Racialization of America. New York: St. Martin's Press.
Young, I. M. 1989. Polity and Group Difference: A Critique of the Idea of Universal Citiz
ship. Ethics 99: 250-74.
Zack, N. 1993. Race and Mixed Race. Philadelphia: Temple University Press.

This content downloaded from 137.158.158.62 on Fri, 31 May 2019 00:41:58 UTC
All use subject to https://about.jstor.org/terms