Received: 6 June 2016 Revised: 21 July 2017 Accepted: 20 September 2017

DO

I: 10.1111/phc3.12468

A R T I C L E

Racial realism I: Are biological races real?

Quayshawn Spencer

University of Pennsylvania

Correspondence

Quayshawn Spencer, University of

Pennsylvania, Department of Philosophy, 249

South 36th Street, Philadelphia, PA 19104,

USA.

Email: [email protected]

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

© 2017 The Author(s) Philosophy Compass © 201

Philosophy Compass. 2018;13:e12468. https://doi.org/10.1111/phc3.12468

Abstract In this article, I discuss and critique how metaphysicians of race

have conceived of and defended racial realism according to how

biologists use “race”. I start by defining “racial realism” in the

broadest accepted way in the metaphysics of race. Next, I summarize

a representative sample of recent attempts from metaphysicians of

race and biologists to defend racial realism and the main criticisms

against each attempt. I discuss how metaphysicians of race have

defended racial realism according to how ordinary people use “race”

in Part II.

1 | INTRODUCTION

In Joshua Glasgow's influential book A Theory of Race, he clearly and succinctly defines “racial realism” as the view that

“race is real,” where “something is real just in case it exists” (Glasgow, 2009, 5).1 While there is variation among

metaphysicians of race with respect to what gets called ‘racial realism’, Glasgow's definition nicely captures the core

of what all metaphysicians of race accept.2 Now, given this thin notion of racial realism, racial realism might seem to be

trivially true. For example, it is widely accepted in the natural sciences that if you can show that even one member of a

kind exists, then that kind exists.3 For instance, that is the strategy chemists use to identify new elements, and it is the

strategy that biologists use to identify new species. Given this assumption, and given that, say, Asians, Blacks, Native

Americans, and Whites are paradigms of races in current American English, is not it trivially true that race itself exists?

Well, it is not that simple.

First, we need to do some metaphysics. To be specific, we need to know what is sufficient for existing in order to

say that Asians, Blacks, Native Americans, and Whites exist. This may sound strange. But think about this. Is not it true

that Tyrannosaurus rex is a paradigm species of the genusTyrannosaurus? But does that fact make it true that the genus

Tyrannosaurus exists? Of course not, because we need to show that the instances of a kind exist in order to use those

instances as evidence that the kind itself exists. Furthermore, showing that anything exists requires adopting some

standards for existing.

Second, we need to do some linguistics. In particular, we need to know what “race” means in the relevant context

in order to say that any particular group exists as a race in that context. This investigation is needed because while it

may be true that Asians, Blacks, Native Americans, and Whites exist, it may turn out to be false that each one is a race.

For example, Blum (2002, 137–146) has argued that Asians, Blacks, Native Americans, and Whites exist, but do not

exist as races given the high bar for what counts as a race in current American English.4 Only after one completes both

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

7 John Wiley & Sons Ltd

wileyonlinelibrary.com/journal/phc3 1 of 13

2 of 13 SPENCER

of these investigations can one use the existence of Asians, Blacks, Native Americans, and Whites, as evidence for the

reality of race.

As you might have guessed, the metaphysical exercise of answering whether race is real has churned out many

different answers. For instance, when the linguistic context of asking whether race is real is restricted to “race” as

understood among biologists (e.g., “race” as a synonym for “subspecies” or “ecotype”), then some philosophers and

biologists have argued that race isn't real.5 However, some philosophers and biologists have argued that race is real

in a biological sense of “race”.6 When the linguistic context of asking whether race is real is restricted to some ordinary

language or dialect (e.g., English, American English, etc.), the most common answer among metaphysicians of race is

that race is real.7 However, more than a few metaphysicians of race have argued that race is not real in the relevant

ordinary sense.8

Now that I′ve discussed how metaphysicians of race define “racial realism,” I'll move on to a discussion of the

major attempts to defend racial realism as “race” is understood among biologists, as well as the major objections to

each attempt in the literature. I will save my discussion of major attempts to defend racial realism as “race” is

understood among ordinary people for Part II.

2 | IS RACE REAL IN ANY BIOLOGICAL USE OF “RACE”?

2.1 | Races as Subspecies

In the last twenty years, there have been five major attempts to defend the reality of race when “race” is understood

in a purely biological way. I will group them into two types of defenses. Some of them are defenses of the reality of

race when “race” is understood as a synonym for “subspecies”. The rest are defenses of the reality of race when “race”

is understood as a synonym for “ecotype”. I will discuss attempts to defend the reality of subspecies first. But before I

begin, I should clarify a few things about subspecies.

In systematics (the study of biodiversity), organisms are sometimes classified into subspecies in addition to

species, genera, and orders. In the most general sense, a subspecies is a subgroup of a species that is an “incipient

species” (Mayr, 1942, 155).9 Moreover, incipient species are groups of organisms that are not quite species, but are

well on their way to becoming species (Mayr, 1969, 44). After Darwinian evolutionary theory was incorporated into

systematics, systematists have felt the need to classify many species into subspecies. This is because Darwin not only

made popular the idea that almost all species originated from evolution, but also that species tend to originate from

gradual evolution due to the dominance of natural selection as the process of speciation (species evolution).10 Thus,

if natural selection is the dominant cause of speciation, we should expect to find that many past and present species

had or have incipient species (a.k.a. subspecies).

In the last 20 years, one major attempt to defend subspecies realism, as I will call it, was Ernst Mayr's defense of

the reality of geographic race.11 According to Mayr (1969, 41; 2002, 90), a “geographic race” or “subspecies” is “an

aggregate of phenotypically similar populations of a species inhabiting a geographic subdivision of the range of

that species and differing taxonomically from other populations of that species.” Mayr (2002, 90) argued that it is

“a biological fact” that geographic races exist in many animal species, and so, from induction, it is “a biological fact”

that the geographic race itself exists. While Mayr (2002, 91) thought that geographic races exist in humans

(Homo sapiens), the case for the existence of geographic races is best made using non‐human animals.

For instance, in a groundbreaking study, Gonder et al. (2011, 4767) identified three phenotypically similar

subgroups of common chimpanzees (Pan troglodytes), each of which occupy a subdivision of the geographic range

of common chimps, and whose genetic variation is large enough to pass conventional tests of “differing

taxonomically”. To be specific, some biologists today use a threshold of σ2a≥0:25 in an analysis of molecular variance

(or “AMOVA”) to test whether the genetic variation among two or more conspecific groups is taxonomically

significant (Templeton, 2013, 263).

SPENCER 3 of 13

An AMOVA is a genetic analysis that divides the total genetic variation in a species (σ2) into genetic variation

among its organisms (σ2c Þ, genetic variation among its local populations of organisms (σ2b), and genetic variation among its groups of local populations (a.k.a. population groups) at a single level of grouping (σ2a); where a local population is a

randomly mating group of organisms. The equation that guides the analysis is below:

σ2 ¼ σ2c þ σ2b þ σ2a (1)

Gonder et al. showed that the total genetic variation among common chimps can be divided in such a way that its

population groups differ taxonomically in virtue of exceeding the σ2a≥0:25 threshold. The AMOVA equation that

Gonder et al. (2011, 4768) found is below:

1 ¼ 0:642 þ 0:057 þ 0:301 (2)

The population groups that Gonder et al. (2011, 4768) used to calculate (2) were Upper Guinea chimps, Gulf of

Guinea chimps, and equatorial Africa chimps. Given thatσ2a ¼ 0:301 and that these three chimp populations exemplify the other essential properties of being a geographic race, it seems unquestionable that geographic races exist in

common chimps, and thus, that the geographic race exists. For a bit of context, the reader might be interested to know

that σ2a ¼ 0:043 for humans when we are subdivided into our most genetically different population groups (Rosenberg, Pritchard, et al., 2002, 2,382).

While the case for geographic race realism might seem airtight, there are some leaks. In particular, some biologists

and philosophers of biology have complained for years that the geographic race is not biologically real because the

threshold used to identify taxonomic difference is biologically arbitrary or subjective.12 For example, there is no

biological theory that justifies picking σ2a≥0:25 as a threshold for subspecific difference. It just as well could have been

0.15, 0.35, or anything else. Furthermore, even if most systematists agree that σ2a≥0:25 is an appropriate threshold to

use, that would give the geographic race a subjective reality at best, as opposed to an “objective” reality (Templeton,

1998, 632–633; Andreasen, 1998, 205).

While this line of attack has been persuasive to many in the literature, I am not convinced. For one, any

quantification of genetic difference in this context is merely a measure of taxonomic difference, not what constitutes

taxonomic difference. In the case of subspecies, what constitutes taxonomic difference is being an incipient species.

So, even if a threshold of genetic difference is biologically arbitrary, that fact does not imply that the geographic race

itself is biologically arbitrary, just that our quantitative tests for subspecific difference are biologically arbitrary. And

they are!

Instead of focusing on the quantity of genetic difference among population groups to measure taxonomic

difference, systematists should look at the quality of genetic difference among population groups. This is because

the genetic difference between two population groups can be very small but biologically important if that difference

is relevant to reproductive compatibility. A good example is the genetic difference among humans and common

chimps. On average, a human's genome differs from a common chimp's genome in only 1.229% of total nucleotides,

however, the small genetic difference between humans and common chimps is enough to sustain a biological species

boundary between the two because humans and common chimps cannot produce viable and fertile offspring

(Yu, Jensen‐Seaman, Chemnick, Ryder, & Li, 2004, 1381).13

In the case of subspecies, a good qualitative genetic test for subspecific difference would be identifying genes for

incomplete hybrid sterility. When two groups can produce viable and fertile hybrids, but viable and fertile hybrids are

limited to a particular type (e.g., a particular sex), that is known as incomplete hybrid sterility. If incomplete hybrid

sterility exists among a set of population groups within a species and the phenomenon has a genetic basis, that is

much better evidence than the current quantitative genetic standard for these groups “differing taxonomically”.

However, a more decisive objection against the biological reality of the geographic race is that geographic races

are not required to be genealogical groups—and by a genealogical group I mean a monophyletic group (or clade), which is

4 of 13 SPENCER

a group containing all and only an ancestor and all of its descendants, or a paraphyletic group, which is a group

containing all and only an ancestor and some, but not all, of its descendants (Wiley & Lieberman, 2011, 9).14 Since

Darwinian evolutionary theory was incorporated into systematics, most systematists agree that real taxa must be

genealogical groups. The reason why is twofold.

First, most systematists have always agreed that “real” or “natural” taxa are the groups of organisms that nature

has produced (Darwin, 1859/1999, 338; Wiley & Lieberman, 2011, 9). Second, most systematists agree with Darwin's

(1859/1999, 109–110) hypothesis in The Origin of Species that the groups of organisms that nature has produced are

genealogical groups, all of which are genealogically connected in a “great Tree of Life”. So, for most systematists today,

to not require taxa to be genealogical groups is tantamount to not requiring taxa to be real. Moreover, we now have

empirical evidence that geographic races sometimes systematically mismatch the genealogical groups that nature has

produced.

For example, in a surprising meta‐analysis, Zink (2004, 561) discovered that 97% of the geographic races that

systematists have identified in birds (Aves) are not monophyletic groups. Also, it might interest the reader to know that

two of the three geographic races that Gonder et al. (2011, 4768) identified in common chimps are neither

monophyletic nor paraphyletic groups (Kaessmann, Wiebe, & Pääbo, 1999, 1161). Due to the worry that geographic

races are not the groups of organisms that nature has produced, the majority of the remaining defenses of subspecies

realism in the past 20 years have advanced genealogical conceptions of subspecies. Below are the views about what

race is from three of the most discussed defenses of subspecies realism in the past 20 years.15

1. “[R]aces” or “subspecies” are “the smallest population units that function as an evolutionary lineage within a

species” (Templeton, 1998, 632; 2013, 263).

2. “Cladistic subspecies” or “cladistic races” are “monophyletic groups” of “breeding populations” in “well‐defined

species” (Andreasen, 1998, 209; Andreasen, 2005, 95–96).

3. “[R]aces” are “inbred lineages” within a species that originated from “populations once geographically separated”

and that “as a result of the separation, there are differences in superficial phenotypic traits,” and “the distinct phe-

notypes” are maintained when conspecifics from different inbred lineages are brought together due to “incipient

isolating mechanisms” that developed during the period of geographical isolation (Kitcher, 1999, 97; 2007, 303).

Due to space constraints, I will not discuss all three of the major defenses of subspecies realism in the past

20 years. Rather, I will just discuss Andreasen's as representative of the group. Andreasen's argument for the reality

of cladistic race is slightly different from Mayr's argument for the reality of geographic race. Remember that Mayr

claimed that geographic races exist in many animal species, and used that claim to infer that the geographic race exists

as well. In contrast, Andreasen (1998, 200) claims that cladistic races “once existed” in the human species, and she

uses that historical fact to infer that the kind cladistic race has “biological reality”.16

Something that's key to appreciating Andreasen's defense for the biological reality of cladistic race is understand-

ing why she thinks that something that satisfies the essential properties of a cladistic race is biologically real. Basically,

she thinks this because she thinks that cladistic races are groups of organisms produced by nature—in this case by the

natural process of evolution—as opposed to merely being a product of “our classifying activities” in systematics—as

she thinks is the case with geographic races (Andreasen, 1998, 207–209). But even so, Andreasen owes us some

empirical evidence that humans once had cladistic races. For that evidence, Andreasen turns to a well‐known

phylogenetic tree first constructed by the population geneticists Luigi Cavalli‐Sforza, Paolo Menozzi, and

Alberto Piazza, which I have presented in Figure 1 below.

While the details of how Cavalli‐Sforza, Menozzi, and Piazza (1994) constructed the tree in Figure 1 are complex,

all you need to know for this discussion is how to interpret the tree and that the tree has been reproduced over and

over again using different background assumptions.17

As for how to interpret the tree in Figure 1, interpret each edge as a breeding population, each internal vertex as a

point in time when a parent population split into two daughter populations, the root as the time when our species

FIGURE 1 A Phylogenetic tree for nine human breeding populations from Cavalli‐Sforza et al. (1994, 80)

SPENCER 5 of 13

originated, and each leaf as a more modern (but unspecified) time.18 Furthermore, interpret each edge incident to a

leaf as representing one of nine breeding populations that Cavalli‐Sforza, Menozzi, & Piazza, 1994, 77–79) have pre-

viously determined to be monophyletic groups of breeding populations. Thus, if Cavalli‐Sforza et al.'s tree is accurate,

the human cladistic races that once existed were the following: New Guineans and Australians, Pacific Islanders

(excluding New Guineans), Southeast Asians, Northeast Asians (excluding Arctic Northeast Asians), Arctic Northeast

Asians, Amerindians, Europeans, non‐European Caucasoids, Africans (excluding North Africans), non‐Africans, and

six other monophyletic groups of breeding populations that I will leave as an exercise for the reader to identify.19

In the literature, there have been three major objections to Andreasen's racial realism. The first is the semantic

mismatch objection. Roughly, this objection states that cladistic subspecies are not races because what constitutes a

cladistic subspecies deviates too much from “the folk concept of race” and “our folk taxonomy” of races (Glasgow,

2003, 459–460).20 For instance, notice that according to Figure 1, neither Asians nor Hispanics (a.k.a. Latinos) were

ever human cladistic races. However, some critics consider these omissions to be a “central flaw” of Andreasen's racial

realism (Glasgow, 2003, 459; Glasgow, 2009, 96). In addition, according to Figure 1, there were many human cladistic

races that ordinary people have never recognized as races. One example is Pacific Islanders, Southeast Asians, and

the parent population of these two. Finally, notice that, according to the concept of a cladistic race, cladistic races

need not be distinguishable by visible phenotypes. However, that is also a fatal flaw of Andreasen's racial realism

for many critics since they see races to be distinguishable by visible phenotypes by definition (Glasgow, 2009, 94

Zack, 2002, 76).

Unfortunately, Andreasen wavered in her response to the semantic mismatch objection. Sometimes she

responded to it by saying there is “quite a bit of agreement” between the human cladistic races that once existed

and some ordinary racial schemes, especially the one used on the 2000 US census questionnaire (Andreasen, 2004,

437). Other times, Andreasen (2005, 105) responded to it by saying that human cladistic races need not match folk

races or folk concepts of race in order to be races because systematists are simply using “race” differently from

ordinary people. The correct response to the semantic mismatch objection is Andreasen's second response.

As Sally Haslanger (2012, 299) has rightly pointed out, there are no “language police” for the word “race” or any

other word. Words get their meanings from linguistic communities conventionally assigning meanings to them, and as

such, any word can have a different meaning across different linguistic communities that intend to use the word

differently. Furthermore, and as I have discussed previously, systematists do intend to use “race” differently from

ordinary people. They intend to use “race” as a synonym for “subspecies”. So, how ordinary people use “race” is

irrelevant to whether Andreasen's defense of racial realism succeeds.

A much stronger objection to Andreasen's racial realism is the treeness objection (Gannett, 2004, 330; Templeton,

2013, 264). Remember that Andreasen relies entirely on Cavalli‐Sforza et al.'s phylogenetic tree as evidence that any

cladistic races ever existed. However, this is a controversial source of evidence because it's controversial that the

evolutionary history of human populations has a tree structure. In graph theory, a tree is a connected graph with no

6 of 13 SPENCER

loops or cycles (Diestel, 2000, 12). Since Cavalli‐Sforza, Menozzi, & Piazza, 1994, 80) represented each human breed-

ing population as a single edge before the time t represented by the leaves, that assumption seems to presuppose that

(T1) each population was reproductively isolated at t and was reproductively isolated at all times from t to when it

evolved (Gannett, 2004, 329). Since Cavalli‐Sforza, Menozzi, & Piazza, 1994, 80) represented the evolutionary rela-

tionships among human breeding populations before t with a tree, that assumption seems to presuppose that (T2)

before t, no population evolved from fusion or hybridization, and every population evolved from branching or anagen-

esis (Gannett, 2004, 329).

Together (T1) and (T2) constitute the necessary and sufficient conditions for an evolutionary history of popula-

tions having a tree structure, and the treeness objection states that human cladistic races have never existed because

the evolutionary history of human breeding populations has never had a tree structure (Gannett, 2004, 330;

Templeton, 2013, 264). Furthermore, if human cladistic races have never existed, then Andreasen lacks the evidence

to say that the cladistic race exists. Also, the critics here do more than make bold claims.

One notable critic, Alan Templeton, devised a test for detecting whether an evolutionary history of populations

has a tree structure—a test called “Multi‐locus nested clade phylogenetic analysis (ML‐NCPA)”—and used that test

on a sample of human breeding populations similar to Cavalli‐Sforza et al.'s sample (Templeton, 2013, 269).21 His

result was that “ML‐NCPA strongly rejects the null hypotheses of no gene flow and no admixture under the null

hypothesis that isolated lineages did exist [in humans]” (Templeton, 2013, 269). In other words, Templeton found that

(T1) is false for every breeding population represented in Figure 1.

While the treeness objection is relevant and formidable, it is not the fatal blow that many race scholars take it to

be against Andreasen's racial realism.22 This is for two reasons. First, it can be true that the evolutionary history of

human breeding populations has never had a tree structure and false that human cladistic races have never existed.

Second, Cavalli‐Sforza et al. do not need to presuppose that the evolutionary history of human breeding populations

has a tree structure in order to reliably use a tree to reconstruct the evolutionary history of human breeding popula-

tions. Both of these claims may sound unbelievable, so let me explain.

Suppose that the evolutionary history of human breeding populations has never had a tree structure. Then, that

implies that there has never been a time when both (T1) and (T2) were true for human breeding populations. But even

if the latter is true, why does that matter for knowing whether human cladistic races ever existed? Andreasen (1998,

209) defines a “breeding population” as “a set of local populations linked to one another by reproductive ties that are,

for the most part, reproductively isolated from other such populations.” Thus, breeding populations can exist without

being completely reproductively isolated from one another. In that case, (T1) being false does not imply that human

breeding populations never existed.

But also, a monophyletic group of breeding populations can exist even if some of its constituent breeding popu-

lations evolved from fusion or hybridization. For instance, some evolutionary biologists take Cavalli‐Sforza et al.'s tree

to be evidence that East Asians evolved from fusion between Northeast (including Arctic Northeast) and Southeast

Asians (Karafet et al., 2001, 625). If that is true, and if Southeast Asian is an extinct breeding population and East Asian

is an extant breeding population, then {Southeast Asian, East Asian} is a monophyletic group of breeding populations

where one population in the group (East Asian) evolved from fusion. So, (T2) being false for human breeding popula-

tions does not imply that monophyletic groups of breeding populations never existed in humans. Provided that H. sapi-

ens is a well‐defined species, all of this shows that it is possible that monophyletic groups of breeding populations in a

well‐defined species (cladistic races) once existed in our species even if the evolutionary history of human breeding

populations never had a tree structure.

Of course, even if human cladistic races could have existed without the evolutionary history of human breeding

populations being a tree structure, the question now becomes how can Andreasen reliably use Cavalli‐Sforza et al.'

s tree as evidence that human cladistic races once existed? The latter is especially perplexing since we now know (from

Templeton's analysis) that the evolutionary history of human breeding populations does not have a tree structure. The

worry here is that since Cavalli‐Sforza et al.'s use of a tree to reconstruct the evolutionary history of human breeding

populations appears to presuppose a tree structure for that history, and since the evolutionary history of human

SPENCER 7 of 13

breeding populations doesn't have a tree structure, it seems to be a mistake to use Cavalli‐Sforza et al.'s tree as

reliably representing the underlying evolutionary history.

However, Andreasen's reliance on Cavalli‐Sforza et al.'s tree to infer that human cladistic races once existed is not

problematic simply because we know that the evolutionary history of human breeding populations does not have a

tree structure. The latter is because Cavalli‐Sforza et al.'s tree is an idealization that is strictly speaking inaccurate,

but accurate enough to be useful in detecting major branchings in the evolutionary history of human populations.23

For instance, the split on Cavalli‐Sforza et al.'s tree that separates Africans from non‐Africans is very robust across

phylogenetic studies and was 98% robust in Cavalli‐Sforza et al.'s bootstrap analysis (Cavalli‐Sforza, Menozzi, &

Piazza, 1994, 80; Cavalli‐Sforza & Feldman, 2003, 269).24 As such, it is not a bad inference at all to say that, at some

point in the past, non‐Africans and Africans were two human breeding populations (in Andreasen's sense) that evolved

via branching from a single parent population. Furthermore, if at least that much is true, then Andreasen can say with

confidence that humans once had at least two cladistic races: Africans and non‐Africans. With that said, the more

recent splits on Cavalli‐Sforza et al.'s tree were less robust in their bootstrap analysis and are less robust across

phylogenetic studies, and so, should be viewed with skepticism.25

Perhaps the strongest objection to Andreasen's racial realism is that she hasn't provided a convincing case for any

cladistic races being biologically real (and thus, real) even if it's true that huma

Related Tags