Выбери формат для чтения
Загружаем конспект в формате pdf
Это займет всего пару минут! А пока ты можешь прочитать работу в формате Word 👇
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/319574776
Kin selection and ethnic group selection
Article in Evolution and Human Behavior · September 2017
DOI: 10.1016/j.evolhumbehav.2017.08.004
CITATIONS
READS
3
166
1 author:
Doug Jones
University of Utah
25 PUBLICATIONS 765 CITATIONS
SEE PROFILE
All content following this page was uploaded by Doug Jones on 20 January 2018.
The user has requested enhancement of the downloaded file.
ENS-06146; No of Pages 10
Evolution and Human Behavior xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Evolution and Human Behavior
journal homepage: www.ehbonline.org
Kin selection and ethnic group selection
Doug Jones
Department of Anthropology, 270 South 1400 East Room 102, University of Utah, Salt Lake City, UT 84112, United States.
a r t i c l e
i n f o
Article history:
Initial receipt 7 December 2016
Received in revised form 7 August 2017
Final revision received 23 August 2017
Available online xxxx
Keywords:
Kin selection
Ethnicity
Ethnocentrism
Cultural group selection
a b s t r a c t
Ethnicity looks something like kinship on a larger scale. The same math can be used to measure genetic similarity
within ethnic/racial groups and relatedness within families. For example, members of the same continental race
are about as related (r = 0.18–0.26) as half-siblings (r = 0.25). However (contrary to some claims) the theory of
kin selection does not apply straightforwardly to ethnicity, because inclusive fitness calculations based on
Hamilton's rule break down when there are complicated social interactions within groups, and/or groups are
large and long-lasting. A more promising approach is a theory of ethnic group selection, a special case of cultural
group selection. An elementary model shows that the genetic assimilation of a socially enforced cultural regime
can promote group solidarity and lead to the regulation of recruitment to groups, and to altruism between
groups, based on genetic similarity – in short, to ethnic nepotism. Several lines of evidence, from historical population genetics and political psychology, are relevant here.
© 2017 Elsevier Inc. All rights reserved.
1. Introduction
1.1. No
The theory of kin selection is a central pillar of the current evolutionary synthesis. The theory is important because it explains the widespread phenomenon of kin altruism – the evolution of behaviors
geared to the survival and reproduction of an individual's kin, at the expense of the individual's own survival and reproduction.
Ethnicity and ethnocentrism in human societies share some affinities with kinship (Connor, 1993; Horowitz, 1985; Weber, 1978). Ethnic
group members often maintain, rightly or wrongly, that they are
descended from a common set of ancestors. They often use the idiom
of kinship for one another – fellow ethnics are “brothers” and “sisters.”
Ethnic identity, like kinship, is commonly seen as a primordial, ascribed,
essential status, not easily changed. And ethnic group relations, like relations among kin, often seem to involve something more – and more
primal – than the rational pursuit of individual or class interests.
All this has suggested to some evolution-minded authors that ethnicity is kinship, and that the evolution of ethnic sentiments can be explained by the theory of kin selection. An ethnic group is an extended
family (so the argument goes), and ethnocentrism is kin altruism, advancing ethnic genetic interests through ethnic nepotism (Harpending,
2002; Rushton, 2005; Salter & Harpending, 2013; Shaw & Wong,
1989; Van Den Berghe, 1981; Vanhanen, 1999; Whitmeyer, 1997). It
would be an important development in social theory if any of this
turned out to be the case. Is this a real possibility? In the next three sections of this paper, I argue that the answer is No, Yes, and Maybe.
One argument for equating ethnicity and kinship is theoretical. The
same mathematical machinery can be used to quantify genetic similarity within individuals and families, and within larger groups ranging
from local subpopulations to continent-scale races. Insofar as ethnic
groups correspond to population subdivisions, the population genetic
definitions of kin relatedness and ethnic group relatedness are the
same, allowing for a change of variables. This equivalence suggests
that – following the theory of kin selection and assuming that ethnic
group relatedness is high enough – we might predict significant altruism within ethnic groups. This possibility is taken up in the next section,
where the verdict is negative. In spite of the formal correspondence,
there is a quantitative difference between families and ethnic groups
that prevents a straightforward application of the theory of kin selection
to ethnicity.
E-mail address: douglas.jones@anthro.utah.edu.
1.2. Yes
The subsequent section arrives at a more positive assessment. It presents an alternative theory in which ethnic nepotism is socially
enforced, and favored by ethnic group selection, a subtype of cultural
group selection. According to the theory, members of an ethnic group
may be cooperative and altruistic toward fellow ethnics based on shared
genes. But shared genes are not just a result of genealogical connections,
as they are in the standard theory of kin selection. Instead, a theory of
ethnic nepotism must take into account some special evolutionary processes at work in human social evolution.
https://doi.org/10.1016/j.evolhumbehav.2017.08.004
1090-5138/© 2017 Elsevier Inc. All rights reserved.
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
2
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
The coefficient of relatedness and the coefficient of inbreeding are
related by the formula
1.3. Maybe
Ethnic group selection is a theoretical possibility; it might or might
not have been of any importance in human evolution. Section 4 briefly
reviews a few pertinent lines of evidence, from historical population genetics and political psychology.
2. From kin groups to ethnic groups
2.1. Relatedness and inbreeding
Hamilton's rule is a simple formula, central to the theory of kin selection (Hamilton, 1964). This section begins with the standard exposition
of the rule, and how relatedness relates to inbreeding. The rest of the
section shows that the rule can be tricky, so that applying it to ethnicity
is not straightforward.
According to Hamilton's rule, an altruistic act that imposes cost cj on
benefactor j, while providing benefit bi for beneficiary i, is favored by
natural selection as long as
c j =bi b r ij ! vi =v j
ð1Þ
Here rij is the coefficient of relatedness: for a gene found in j, if k is the
expected number of copies of the gene in j, then rij·k is the expected
number of copies in i. If j is not inbred, then k = 1. If j is inbred, then k
N 1. This counts only genes identical by descent over and above the
genes i and j share as members of the same population. This version of
Hamilton's rule also includes terms vi and vj, the reproductive value of recipient and donor, i.e. their expected genetic contribution to distant future generations. This might depend on their ages; we see in the next
section why this matters.
The coefficient of relatedness is connected to another quantity, the
coefficient of inbreeding, F (Falconer & McKay, 1996; Frank, 1998).
The coefficient of inbreeding is, in the first instance, a measure of genetic
similarity within a diploid individual, the probability that maternally
and paternally inherited copies of a gene are identical by descent. We
can write this as Fjj for individual j. The coefficient is greater than zero
if j's mother and father are related. For example if j's parents are sister
and brother, then Fjj = 0.125. Once again, this is over and above the
probability that maternal and paternal copies are the same just because
mother and father are members of the same population.
The coefficient of inbreeding can also be used to quantify genetic
similarity within a subpopulation that is part of a larger population.
This is usually written FST. If individuals tend to find mates in their
own subpopulation, but to mate randomly within their subpopulation,
then the probability Fjj that maternal and paternal copies of a gene in individual j are identical by descent is equal to the probability Fij = FST that
two genes in randomly selected individuals i and j in the subpopulation
are identical by descent.
!
"
r ij ¼ 2 ! F ij = 1 þ F ij
ð2Þ
where the 2·Fij term takes into account that i, being diploid, has two
chances of having genes identical by descent with a gene in j, and the
1 + Fjj term takes into account that at homologous loci j may be identical by descent with herself through inbreeding.
Various authors have been interested in how coefficients of inbreeding and relatedness might relate to the evolution of human social behavior in groups larger than families. Some of their results are shown
in Table 1, which gives summary statistics for FST's for assorted human
population subdivisions, as well as the corresponding coefficients of relatedness (column headed rH) following Condition (2) with Fij = Fjj =
FST.
It is tempting to plug the rH values in the table into Hamilton's rule,
and predict kin altruism accordingly. Several of the authors cited in the
table have done just this, reaching different conclusions depending on
what level of population subdivision they think is evolutionarily important (Bell, Richerson, & McElreath, 2009; Harpending, 2002; Salter &
Harpending, 2013).
We'll see below that things are not so simple.
2.2. Kin selection: socially enforced altruism
The simplest formulation of the theory of kin selection treats it as a
one-player game, where an actor has the power to help one or more
passive recipients. In this case (given some further assumptions; see
below) the r's derived from genealogies or from across the whole genome may predict behavior toward kin.
But the theory gets more complicated when there are strategic interactions between players. For example, imagine a game, in the game theory sense, played by two siblings. If the only thing one player knows
about the other is that he is her brother, then she can expect that half
his genes are identical by descent with hers. But if she also knows
what strategy her brother has chosen, then this may raise or lower the
estimated number of shared genes at loci affecting the choice of strategy
(but not at other unlinked loci). It will be adaptive for her to raise or
lower her level of altruism accordingly. In a case like this, neither genealogy nor genome-wide genetic similarity suffices to predict similarity
at loci governing strategic behavior, and it is these loci that kin selection
cares about. So one way the theory of kin selection gets tricky is when
it's combined with game theory. Just assuming that game players keep
score according to Hamilton's rule, with r's based on genealogy, generally gives the wrong answer.
To some extent, each case that combines kin selection and game theory has to be analyzed separately. But there is a family of cases that can
be treated more systematically – if sometimes approximately – involving socially enforced nepotism (Jones, 2000, 2016). Socially enforced nepotism happens when a group of individuals acts together to help
another related group, without much or any expected return benefit.
Table 1
Inbreeding and relatedness: summary statistics.
Study
Type of society or population subdivision
Number of populations
Subdivision size
Median
(range)
FST
Median
(range)
rH
Median
(range)
rG
Median
(range)
Jones (2000)
Tribal populations
10
Foragers
13
Bell et al. (2009)
Adjacent nations
59 pairs
N105
Salter and Harpending (2013)
Races
1 (Homo sapiens)
N108
0.030
(0.003–0.063)
0.076
(0.007–0.170)
0.0032
(0.032–0.00044)
0.12
(0.10–0.15)
0.058
(0.006–0.119)
0.141
(0.014–0.29)
0.0064
(0.063–0.00088)
0.22
(0.18–0.26)
0.822
(0.231–0.991)
Bowles (2006)
1875
(500–122,022)
–
1.00
1.00
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
3
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
(This is phrased as one group helping another group, but socially
enforced nepotism can also involve a well-off segment of a kin group
collectively enforcing a rule of helping needy group members who cannot reciprocate.) Socially enforced nepotism is distinct both from individual nepotism, where individuals act on their own to help kin, and
from simple cooperation, where members of a group act together to
their mutual advantage.
The idea behind socially enforced nepotism is that altruism toward
kin is a public good. When someone pays a cost to provide a benefit
for one of her kin, she is also providing a free inclusive fitness boost to
the rest of the recipient's kin. If individuals act together to help their
mutual kin, then natural selection favors a higher level of altruism, measured as a cost/benefit ratio, than if they act separately. Standard inclusive fitness calculations aren't guaranteed to give the right answers here
(see, for example, the discussion of the Brothers Karamazov Game in
Jones, 2016).
Socially enforced nepotism, like other public goods provisioning, requires enforcement. Several enforcement mechanisms have been proposed. Bowles (2006) argues that reproductive leveling in foraging
societies amplifies kin altruism. Jones (2016) argues that reputation
can operate as a kind of social currency, allowing kin groups to enforce
an ethic of generalized reciprocity, in which needy kin get help even
though they can give back little in return.
One way to get a handle on socially enforced nepotism is to recognize a group coefficient of relatedness, rG, in addition to the more familiar individual coefficient of relatedness, rH. The group coefficient of
relatedness determines the expected level of altruism of members of a
group according to the formula c/b b rG, provided the members of a
group act together. More precisely, rH gives the exact break-even altruism ratio, c/b, provided that each group member has one vote regarding
the amount of altruism carried out by the group, and the amount increases linearly with the number of Yes votes. But the formula is still approximately correct even with different voting rules and enforcement
mechanisms (Jones, 2000).
Table 2 compares F, rH, and rG. Each of these variables is a coefficient
of relatedness at some level, whether between genes (F), between diploid individuals (rH), or between groups (rG). And each variable is a regression coefficient. The expected number of genes identical by
descent in a target unit – gene, individual, or group – equals the number
of such genes in a focal unit, times the appropriate coefficient.
The right-most column of Table 2 shows how adjacent levels of relatedness are connected. Each equation there supplies a kind of “voter's
guide.” The first equation tells a gene how much individual altruism to
“vote” for when it finds itself inside a diploid organism. The second
equation tells an individual how much socially enforced nepotism to
vote for when she finds herself inside a collectively acting group.
We'll see below that socially enforced altruism is also relevant to
ethnicity. But first we take a detour.
2.3. Kin selection: the weak selection assumption
Strategic interaction is not the only factor that can throw off
Hamilton's rule. Even if we ignore strategic interaction and consider
only individual rH's, we arrive at some odd conclusions if we're not
careful.
Consider, for example, F and rH for major continent-scale races (the
last row in Table 1). The F values demonstrate a widely publicized result: members of different races are more alike than different
(Lewontin, 1972). There is far more genetic variation within races
than among them (90–85% versus 10–15% of total variance). The same
figures, converted into rH's, also demonstrate a less familiar result:
members of the same race, relative to the species as a whole, are related
to one another (rH = 0.18–0.26) almost as closely as half-siblings (rH =
0.25).
If we were to plug an rH in this range into Hamilton's rule, we would
predict high levels of altruism within major population blocs. Is this remotely plausible? Are random strangers of the same race in a multiracial
society nearly as spontaneously altruistic to one another as a pair of half
siblings, a grandparent and grandchild, or an uncle and niece in a racially homogenous society? Surely something has gone wrong here; standard kin selection theory is being misapplied somehow. It's not
enough to claim that multiracial societies are an evolutionary novelty,
so that human beings haven't had enough time to evolve the necessary
adaptations. The fallacy runs deeper; even if humans had evolved in a
multiracial setting, standard kin selection would not favor this sort of altruism. It's not that ethnic groups and races are categorically distinct
from families. The same variables – coefficients of inbreeding and relatedness – can be used to quantify genetic variation both for small
ephemeral groups and for large enduring ones. But there is a crucial
quantitative difference between small and large groups. With increasing
scale and time depth come increasingly serious violations of one of the
assumptions made in deriving Hamilton's rule, the weak selection
assumption.
According to the weak selection assumption, selection doesn't
change the frequency of an altruism gene between the time it is present
in any shared ancestors and the time it expresses itself in a descendant's
altruistic act. We can see why this matters by dropping the assumption.
Suppose, for example, that a woman has a chance to save the life of her
full brother. Under the standard argument invoking Hamilton's rule,
there is a probability of 0.5 that an altruism gene found in the rescuer
is found in her brother as a result of their shared parentage, so she
should save him, even at some risk to her own life, as long as his expected fitness gain, times 0.5, is greater than her expected loss. But now suppose that some of the potential rescuer's siblings have already died
while altruistically rescuing some of her other siblings. Their deaths
will have removed copies of the altruism gene from the family. The expected frequency of the allele among the survivors, including her imperiled brother, is now b 0.5.
More realistically, selection doesn't usually change gene frequencies
a lot in such a short period. For groups lasting several generations, up to
the scale of small local kin groups, the weak selection assumption is likely roughly correct (Bowles, 2006). But over longer time scales, the assumption becomes increasingly unrealistic. Even slight fitness
differences can accumulate to cause large changes in gene frequencies
in long-lived groups. This is crucial with respect to the larger population
subdivisions (nations and races) in Table 1, because the F values given in
the table omit one key detail. These F's are valid for most genetic loci,
which are not under strong selection. But for genes under selection, F's
can be much lower or higher. For example, for genes governing pigmentation, directional selection has pushed populations much farther apart
Table 2
Inbreeding and relatedness: definitions and formulas.
Variable symbol
Variable name
Measures relatedness of …
F
rH
Coefficient of inbreeding
Individual coefficient of relatedness
One gene to another, within an individual, between kin,or within a subpopulation.
One diploid individual, i, to another, j.
rG
Group coefficient of relatedness
One group, i, to another, j, where
rH within j = rjj
rH between i and j = rij
and n's are group sizes
Formula
2F
r H ¼ 1þ Fijii
n !r
ij
r G ¼ 1þðnij −1Þr
ii
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
4
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
than drift alone would have carried them (Sturm, 2009). This means
that in really long-lived groups, the theory of kin selection only applies
if selection is very weak. Very weak selection takes a very long time –
many cycles of group formation and replacement – to change gene frequencies appreciably. And the end result – tiny altruistic benefits, tinier
costs – is trifling.
This criticism, I think, is fatal to Salter and Harpending's (2013) case
for ethnic nepotism. And the criticism applies in reverse to Bell et al.
(2009). These authors argue that where the evolution of large-scale
prosociality is concerned, we should be looking at r's between adjacent
nations, rather than among continent-scale races, as in Salter and
Harpending. Because these r's are low (they write), we can't expect
much kin-based altruism at the level of ethnic groups or nations. But I
argue instead that this is beside the point. Suppose the r's between nations had been higher: this, on its own, would not license a prediction
about kin altruism among ethnic groups. On a large scale, for nontrivial benefits and costs, neither set of r's can be plugged into
Hamilton's rule.
Other authors have criticized the argument that kin selection theory
predicts ethnocentrism (Brigant, 2001; Dawkins, 1979). Here I have
tried to explain both why the argument has appealed to some theoretically adept scholars and why it fails.
3. From cultural group selection to ethnic group selection
The standard theory of kin selection does not yield a theory of ethnic
nepotism. In large, long-lasting groups, members may or may not have
high coefficients of inbreeding and relatedness as a result of sharing
multiple links to distant ancestors, but neither genealogical relatedness
nor genetic similarity over the whole genome is a reliable guide to similarities at loci governing sociality. However, the fundamental insight
behind the theory of kin selection – that natural selection can favor social interaction, including altruism, based on shared genes – could still
be relevant even on the scale of ethnic groups, where shared genes
may have more to do with the selection pressures associated with
shared culture. This section develops a model to illustrate this.
The model here builds on previous work on cultural transmission
and social evolution, and on the extensive literature (Richerson et al.,
2016) making a theoretical and empirical case for cultural group selection. The model takes this work in a new direction, concentrating on
what distinguishes ethnicity from large-scale cooperation in general,
and on the parallels between ethnicity and kinship. It can be regarded
as a follow up to my previous work on kinship norms and evolution.
In that work I suggested – following a long-standing tradition in social
anthropology – that what distinguishes kinship norms from other social
norms is an ethic of unbalanced generalized reciprocity among kin, and I
developed a model of kin altruism amplified by socially enforcement
that might account for this. The present work is also about socially
enforced altruism, but on a larger scale, where the weak selection assumption no longer holds. In contrast with much previous work, the
focus is less on the sociological details of how norms are enforced, and
more on the consequences of enforcement for evolution within and between groups. The result is a simple model that makes explicit some
parallels between ethnic group selection and kin selection which are
largely unexamined in earlier work. For example, we see that ethnic genetic relatedness, ethnic nepotism, and ethnic genetic interests can all
be meaningful concepts.
We begin with a gene-free model of cultural group selection
(Subsection 3.1). This is elaborated to distinguish ethnic group selection
as a specific subtype of cultural group selection, resulting from the genetic assimilation of culture (Subsection 3.2). The remaining subsections highlight some contrasts between cultural group selection in
general, and ethnic group selection. With cultural group selection,
groups may expand either through natural increase or through
recruiting outsiders. With ethnic group selection, genes make a difference, and recruitment (e.g. through exogamy) may be favored or
disfavored depending on the genetic makeup of potential recruits, and
the strength of selection producing genetic similarity within groups, potentially leading to policing of genetic boundaries between groups
(Subsection 3.3).
Also, while most models of cultural group selection are meant to account for cooperation, with members of a group paying a cost to secure a
collective good for all, the present model also covers collective altruism,
with members of one segment of an ethnic group paying a cost to secure
benefits for another segment (Subsection 3.4). This is socially enforced
altruism. It depends on shared genes, and we define an ethnic coefficient of relatedness, rE, between different segments of an ethnic
group, comparable to the r's discussed above, but depending in this
case on a segment's history of selection. This coefficient is determined
by the ratio of altruism genes in donor and recipient. And it determines
the ratio of donor costs to recipient benefits where altruism is concerned, although in a slightly complicated fashion, since the coefficient
evolves over time.
The theory of ethnic group selection developed here does not cover
everything falling under the heading of ethnicity. Some evolutionary
models of ethnicity are concerned with trust and cooperation in dyads
without ethnic group selection (Hartshorn, Kaznatcheev, & Schultz,
2013; McElreath, Boyd, & Richerson, 2003). Nor is the theory a general
theory of social evolution. For example, much of the evolution of
major missionary religions involves cultural group selection without
ethnic group selection (Norenzayan, 2015). But the theory may provide
insight into a range of phenomena, noted by scholars of ethnicity and
ethnonationalism, that look something like kin altruism: human beings
often identify with ethnic groups, that, in contrast to other social groupings, are defined by putative common descent. And people often organize to support the supposed interests of those groups, police group
boundaries, and enforce altruism toward group members. The theory
may also tie together some findings regarding the psychology of ethnocentrism, including individual-level variation (see especially Subsection
4.2).
3.1. The selection/defection balance
We begin with an elementary gene-free model of cultural group selection for provision of public goods. Consider a population of groups. A
group may be in one of two cultural regimes, high or low solidarity, labeled U and V. In the high solidarity regime, U, each group member pays
a cost c to make a contribution to a public good which results in benefit b
for each. In the low solidarity regime V, group members neither pay the
cost nor gain the benefit. Costs and benefits are measured in the currency of fitness, and b N c, so in regime U group members have fitnesses
proportional to 1 + su with su = b − c. In regime V, group members
have fitnesses proportional to 1.
With genetic evolution, natural selection in favor of an allele may be
balanced by mutation pressure in the opposite direction. A similar balance may hold with cultural evolution. Suppose groups sometimes
spontaneously transition from one cultural regime to the other. In
each generation, a U group switches to V with probability f, and a V
group switches to U with probability h.
For given values of su, f, and h, the equilibrium frequency, û, of U is a
solution to a quadratic equation. But here we use an approximation. We
are most interested in the balance between cultural selection in favor of
high solidarity (given by su), and cultural defection away from it (given
by f). We assume that solidarity is prone to spontaneous decay, so su and
f are the important variables. The opposite transition, from low to high
solidarity (given by h), is infrequent. With h close to 0, û is approximately given by:
^ ≈ 1−f =su
u
if
^ ≈ 0 if f N su
u
f ≤su
ð3Þ
ð4Þ
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
This is similar to the standard formula for mutation-selection balance in genetic evolution.
We can refine the model by considering evolution with more than
one solidary regime. Suppose that in addition to the non-solidary regime V, there are two possible solidary regimes, U1 and U2, with associated frequencies u1 and u2 (where u1 + u2 + v = 1), selection
coefficients su1 and su2, and defection rates f1 and f2. Suppose f1/su1 N
f2/su2. Then the growth rate for U1 is positive only up to u1 + u2 b 1 −
f1/su1, while the growth rate for U2 is positive all the way to u1 + u2
b 1 − f2/su2, so there is an interval where the rate is negative for U1
and positive for U2. In consequence, U2, the solidary regime with a
lower f/sU, competitively excludes U1. So as long as h is close to 0, cultural group selection favors the regime with the smallest f/sU, in other
words, with a combination of high fecundity (high sU) and high transmission fidelity (low f).
3.2. Genetic assimilation
In the standard theory of kin selection, gene frequencies differ between groups as a result of sampling error. By chance, some families
have more altruism alleles, others fewer. But in long lasting groups,
within-group selection can be a more important influence on gene frequencies. Within-group selection can either undermine or strengthen
selection on culture among groups.
Let us return to our population of groups under two cultural regimes,
U and V. Now suppose that there are also two alleles, P and Q at one genetic locus. Suppose that in groups under regime U, the frequencies of P
and Q are p and q = 1 − p. And suppose that P is favored by selection
within U groups, so the fitness of P within the group is proportional to
1 + sp., with sp. N 0, while the fitness of Q is proportional to 1. As before,
we are mostly interested in regime U. For regime V, we assume merely
that transitions from V to U are rare (h ≈ 0), and that selection within V
groups keeps their frequency of P uniformly low.
Now suppose that in U groups, P is not only favored by selection, but
also changes the fidelity of cultural transmission. In an all-Q group, the
rate of defection from U to V is f, as before. But in an all-P group the
rate is g. The relationship is linear, so in mixed groups the rate of defection is g·p + f·q.
Under these assumptions, P goes close to fixation in groups under regime U, and the new approximate equilibrium frequency of regime U is:
^ ≈ 1−g=su
u
if g ≤su
^ ≈ 0 if g N su
u
ð5Þ
ð6Þ
With regard to the effect of within-group selection on the evolution
of solidarity, there are two cases to consider:
f −g b 0
f −g N 0
Discordant
Concordant
ð7Þ
ð8Þ
First take the discordant case. Suppose that in mixed groups, P manages to shift some of the cost of producing public goods onto Q. The benefit, b, of belonging to the group is still the same, as is the average cost, c,
but for a carrier of allele P the cost is just c − q·sp., while for a carrier of
Q, the cost is c + p·sp. where 0 b sp. b c. At the same time, with g N f,
groups with higher p are more likely to give up on group solidarity altogether. So selection within groups in favor of P, the reluctant cooperator,
also results in an increased rate of defection and a lower equilibrium u.
Within-group selection could even result in high group solidarity largely disappearing. If g N sp. N f, then the occasional groups that switch from
V to U, beginning with a low frequency of P, will flourish for a while
under cultural group selection, but will be overwhelmed by defection
as group p‘s increase.
5
So far, we seem to have taken a roundabout route to discover that
genetic altruism can't persist in groups much larger than families. This
conclusion holds in large, long-lasting groups even with high values of
F and rH for selectively neutral genes, because selection erases
between-group differences for altruism genes. Even a group that starts
out by chance with an especially low frequency of the uncooperative P
allele still winds up with a high frequency of P after generations of
selection.
But now take the concordant case. If straight altruism can't evolve by
group selection, socially enforced cooperation is another matter. Suppose we return to a population of groups under two cultural regimes,
U and V, with two alleles, P and Q. We add social enforcement: the
cost of solidarity, c, is the sum of two components: the cost, c1, of contributing to public goods, and the cost, c2, of enforcing a contribution.
Total costs are still less than benefits, sU N 0, where sU = b − c. But
this time we make P more committed to group solidarity, and assume
that in mixed groups, regime U rewards the zealous P′s with a smaller
share of the enforcement cost, c2 − q·sp., and punishes the halfhearted Q's with a larger share, c2 − p·sp., where 0 b sp. b c2. At the
same time, g b f, so groups with higher p are also less likely to give up
on cooperation. Selection within groups in favor of P, the committed cooperator, also results in a lower rate of defection and a higher equilibrium u.
When genic natural selection for social enforcement is added to cultural group selection for solidarity, group solidarity can turn from rare to
commonplace. If f N su N g, and h is close to 0, then groups will occasionally switch from low to high solidarity. With f N s, defection initially outweighs cultural group selection, and most of these groups defect from
the high solidarity regime. But occasionally, by chance, a group persists
in regime U for some time. In this case, P increases in frequency, asymptotically approaching p = 1. High frequencies of P stabilize solidarity in
these groups, and they increase their share of the population, up to the
point that cultural group selection, su, is balanced by the lower rate of
defection, g.
We have described two different scenarios, in which natural selection within groups either undermines or reinforces cultural
group selection. If enforcement costs are not too great, then groups
that maintain the latter condition will prevail. Suppose two
solidary regimes U 1 and U 2 have associated selection coefficients
s 1 = b − c 1 and s 2 = b − c 1 − c 2 , where c 2 is the enforcement
cost. Regime U 1, which entails lower costs and a higher betweengroup selection coefficient, has the advantage initially. But U 2
wins ultimately if selection within groups leads to a sufficiently
low defection rate for U2 relative to U1 , i.e. if g 1 /s1 N g 2/s2, in spite
of s 1 N s 2 , where g 1 and g 2 are the respective defection rates after
within-group selection.
What we have outlined here is a group-level version of a familiar
evolutionary phenomenon, genetic assimilation (Crispo, 2007;
Ehrenreich & Pfennig, 2015). In the standard individual-level version, an organism produces a novel adaptive behavior as a result of
learning or other developmental plasticity. Over time, natural selection favors genes that make it easier to acquire the behavior, so the
adaptive behavior is produced consistently rather than sporadically.
Eventually a learned adaptation becomes instinctive, not through Lamarckian evolution, but through natural selection. Cultural transmission is a special form of developmental plasticity, and cultural
group selection can result in a special kind of genetic assimilation,
in favor of group solidarity.
In short: A cultural regime is evolutionarily discordant if sets up
selection pressures against those who keep the regime going. It is
concordant if it does the opposite. Within the limits set by enforcement costs, the combination of cultural group selection and genic
within-group selection favors solidary, concordant groups. This
combination we can call ethnic group selection. The greater the contribution of genetic assimilation to the success of a culture, the greater the “ethnic” component to cultural group selection.
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
6
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
3.3. Recruitment
Apart from selection, another process that can change gene frequencies is migration (Rogers, 1990). In some group selection models, migration is highly inimical to altruism. Even extremely low rates of migration
– the successful emigration, from a group containing non-altruists, of
more than one non-altruist individual over the whole lifetime of the
group – can be enough to eliminate altruist genes. This implies that
group selection rarely favors the evolution of individual altruism. But
the consequences of migration are different when cultural group selection and socially enforced cooperation are involved.
In the case of cultural transmission, the equivalent of migration involves not just physical transfers between groups, but recruitment. A
group may expand by taking in new members who adopt the local culture. At the same time, the recruitment of outsiders can change a
group's genetic composition.
Let's return to our elementary model, a collection of groups under
regimes U and V. The frequency of P, a concordant high-solidarity
gene, in U is p. A U group recruits members from other groups, including
some V groups. The frequency of allele P among these recruits is pm = 1
− qm. In this case, the equilibrium u, for u N 0 is given by:
^ ≈ 1−
u
g þ pd ð f −g Þm=sp
su þ m
ð9Þ
There are two new variables here, pd. and m. The variable pd.
measures the deficit in P among recruits compared to group members,
pd = 1 − pm/p. The value of pd. depends on the frequency of U and V
groups, and on the frequency of P and Q under each regime, which
can be calculated numerically for particular cases.
The variable m is the fraction of the membership of a U group that is
recruited from outside in each generation. The variable shows up twice
in the equation. In the denominator, m shows up along with su, indicating that a group can grow both by natural increase (biological reproduction, the su term) and by recruitment (social reproduction, the m term).
The variable m also shows up in the numerator, indicating that gene
flow from outside increases the rate of U-to-V defection. This effect is
stronger if pd.(f − g) is large, so that recruits have a high frequency of
Q alleles and/or Q raises defection rates a lot. The effect is also stronger
if sp. is small, so that within-group selection is not very effective in removing the Q alleles introduced by recruitment.
The double role of m in Condition (9) implies that recruitment entails another fecundity/fidelity tradeoff. Recruitment increases the (cultural) fecundity of groups, but reduces the fidelity with which culture is
transmitted to the next generation. The break-even point, where dû/dm
= 0, is given by g/su = pd.(f − g) / sp. Below this point, an increase in m
means an increase in equilibrium u. Above this point, an increase in m
means a decrease in equilibrium u. The critical point marks the dividing
line between two selective domains:
g pd ð f −g Þ
N
su
sp
Open
ð10Þ
g pd ð f −g Þ
b
su
sp
Closed
ð11Þ
In the preceding subsection, we saw that selection among and within groups combined – ethnic group selection – can favor solidary concordant groups. In this subsection, we encountered two more
variables, pd. and m, that ethnic group selection might act on. Either variable could be under group control, and might be dialed up or down to
promote group reproduction, although, as before, we must allow for enforcement costs and other tradeoffs. First, a group might change pd., by
changing its sources of recruitment. Other things being equal, groups
that minimize pd will be most successful. Second, a group might change,
m, its rate of recruitment. This variable plays an equivocal role. In the
open domain, the most successful groups produce as many cultural offspring as possible, by a combination of natural increase and the recruitment of outsiders. In the closed domain, the most successful groups
expand through natural increase, and avoid recruiting outsiders, who
raise the probability of defection.
3.4. The problem of altruism
We turn from migration to another area where the genetic makeup
of groups makes a difference.
Cultural group selection can favor group cooperation, where all group
members pay the cost of a public good and all benefit. But it can also favor
socially enforced altruism, directed from one group to another. (We already met with socially enforced altruism above, in the form of socially
enforced nepotism in small kin groups.) Considered at the individual
level, socially enforced altruism doesn't look like altruism: a helper is motivated by social rewards and punishments rather than by altruistic sentiments toward recipients. Considered at the group level, however, socially
enforced altruism really is altruism: a group is pushing its members to
pay a fitness cost to help members of another group. Socially enforced altruism between groups – whether ethnic groups or kin groups – is the
group analog of altruism between individuals.
Imagine once more a population subdivided into groups. Sometimes
opportunities arise for one group to help another, with the helping
group paying a fitness cost c to provide a benefit b for the recipient
group. (These are total costs and benefits, not costs and benefits per
capita.) This might involve the donor group inviting the recipient
group to share in its common property. Or the donor group might enforce a moral code that rewards its members for acts of kindness to
members of the recipient group. Or the donor group might ally militarily
with the recipient group, instead of with a culturally unaffiliated group.
As before, each group belongs to one of two cultural regimes, U or V. In
this case, U groups help other, needy U groups when the opportunity
arises, while V groups provide no assistance. In the simple gene-free
case, regime U is favored by cultural group selection as long as b N c, so
that u increases up to the point that selection is balanced by defection.
For this to work, altruism must be discriminating: U groups must direct assistance selectively to other U groups, rather than to V groups.
This in turn poses the problem of how one U group recognizes another.
One possibility is that a group which is known to have helped other U
groups in the past is recognized as a U group in good standing, and
deemed worthy of assistance when in need. This makes altruism between groups a multi-generation, group-level version of indirect reciprocity (Leimar & Hammerstein, 2001; Nowak & Sigmund, 1998).
Briefly, indirect reciprocity is an extension of direct reciprocity. In direct
reciprocity, you help those who have helped you. In indirect reciprocity,
you maintain a reputation as someone deserving of help by helping
others who are deserving of help. Both theoretical and empirical work
has demonstrated that indirect reciprocity can operate between individuals. If we accept that a group can impose a code of conduct on its
members, then indirect reciprocity among groups is a logical extension
of indirect reciprocity among individuals. But so far this is not much like
kin altruism. Groups help other groups with similar phenotypes –
helpers help helpers – not similar genotypes.
Suppose, however, we again add genes to the model, in the form of
alleles P and Q, with P being positively selected within U and, concordantly, lowering the rate of defection from U. Let different U groups
have different frequencies of P and Q. Then they also have different likelihoods of leaving descendant U groups, according to:
ð f −gÞ=sp
wi ¼ pi
ð12Þ
(See Appendix.) Here wi is proportional to the expected long-term
number of descendant U groups produced by group i, where i is in regime U and has a frequency of P equal to pi. If pi is high, then so is wi,
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
7
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
≡ wi =w j
ð13Þ
where rE = pi / pj.
Condition (13) looks something like Condition (1), the version of
Hamilton's rule given near the beginning of this article. We can compare
two kinds of altruism, individual altruism resulting from kin selection,
and socially enforced altruism resulting from ethnic group selection,
by comparing these conditions.
According to the kin selection condition, the “value” of one individual to another is the product of two components, rij and vii/vj. There is
a separation of variables here, between the coefficient of relatedness
and the ratio of reproductive values. This separation is possible because
the coefficient of relatedness of i to j depends on their genealogical past,
while the reproductive values of i and j depend on their expected reproductive futures, as determined by life history variables like age.
In contrast, in Condition (13) the “value” of group i to group j is
folded into a single component, expressed in terms of either rE or wi/
wj. We can call rE the ethnic coefficient of relatedness. Like other r's, it is
a regression coefficient, the ratio of expected numbers of altruism
genes in recipient and donorj. Alternatively, the ratio wi/wj. is comparable to the ratio of reproductive values in Condition (1). The equivalent rE
and wi/wj terms look both to the past – because the genetic makeup of a
group depends on its history – and to the future – because the genetic
makeup of a group, evolving over time depending on (f − g) / sp., determines the fidelity of cultural transmission. This has no parallel in the kin
selection case, where the mutation rate is assumed to be negligible, and
in any case doesn't vary with r.
Another important contrast between kin selection and ethnic group
selection shows up if we think about pi and pj in Condition (12). Both
kinds of selection depend on groups being genetically different at loci
involved in altruism. But the evolutionary forces that generate group
differences are not the same. With kin selection, between-group differences result from random variation in gene frequencies between families. But with ethnic group selection, between-group differences in p
may result from a variety of evolutionary processes. Random genetic
drift is one of these, but in large long-lasting groups, selection and migration can be more important.
And this in turn entails a further contrast between kin selection and
ethnic group selection: because the sources of genetic variation among
groups are different in the two cases, the cues to relatedness will be different as well. Specifically, with ethnic group selection, some of the important cues may involve a group's history of selection and recruitment.
For example, selection may result in groups with the same “phenotype”
(belonging to the same cultural regime) having very different “genotypes” (different gene frequencies). The length of time a group has
spent in a regime is a cue to its genotype. A group in regime U can figure
out the frequency of P in another U group if it knows how long the other
group has been in regime U, according to the formula:
p½t ' ¼
p0 ! esp !t
1−p0 þ p0 ! esp !t
4. Ethnicity and evidence
There is one further difference between kin selection and ethnic group
selection. Genetic similarity within families – the basic requirement for
kin selection – follows automatically from the way sexual reproduction
works. By contrast, genetic similarity within ethnic groups at relevant
loci – the basic requirement for ethnic group selection – depends on
special conditions that might or might not have held in the evolutionary
past. So ethnic group selection is an iffier proposition than kin selection.
Below we consider whether it is a realistic possibility, in light of several
lines of evidence. The discussion is necessarily brief. Other important
lines of evidence, including work in political science and history on
ethnicity and ethnonationalism, must be left for another occasion.
4.1. Population genetics and population replacement
In the model developed here, some ethnic groups replace others.
Group replacement may explain some puzzling features of human
a
Space
ð f −g Þ=sp
c j =bi b r E
if defecting U groups, by “seeding” regime V with altruism genes, increase the rate of counter-defection from V to U.
Here, however, we keep things simple, in order to focus on comparing kin selection and ethnic group selection. Up to this point, the comparison has emphasized how they differ; ethnic group selection is not
kin selection. Yet the two processes also share important similarities.
Both involve more than just self-interested cooperation; they are
about altruism based on shared genes. More generally, as a result of ethnic group selection, members of ethnic groups may act jointly to regulate sources and rates of group recruitment, and to help fellow ethnics,
guided by cues to ethnic group genetic relatedness. This would involve
group members establishing and enforcing norms and institutions. I
suggest we might reasonably call this ethnic nepotism.
Time
b
Space
and group i will lose fewer descendants through defection, and contribute more descendant groups to future generations, than a group with
low p.
This in turn has implications for the evolution of socially enforced altruism between groups. Whether group j should pay cost cj to provide
benefit bi to group i (where both groups are in regime U) will depend
on wi and wj, according to
ð14Þ
where t is the time spent in regime U, and p0 is the frequency of P when
a group first switches from V to U. Altruism toward the other group can
then be modulated accordingly.
This example could obviously be elaborated. For example, we could
explore cues involving different histories of recruitment. Or we could investigate the indirect route to cultural/genetic reproduction that occurs
Time
Fig. 1. a. Population bottleneck. b. Population replacement.
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
8
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
population genetics (Premo & Hublin, 2009). Humans display less genetic diversity than chimpanzees or gorillas; the past effective population size for humans (around 104) is considerably smaller than for
great apes. This is odd, because humans have had a greater geographic
range, and probably higher total population, for some time.
The usual interpretation of this finding is that human beings passed
through a bottleneck in which total population was small (Fig. 1.a). In
the most dramatic bottleneck scenario, humanity came close to extinction with the Toba volcanic eruption 74,000 years ago, and rebounded
from a small population of survivors (Ambrose, 1998).
There are problems with any model involving a population bottleneck (Harpending et al., 1998). Different genetic loci give inconsistent
estimates of effective population sizes and timing of bottlenecks. Furthermore, small effective population sizes are characteristic not only
of modern humans, but of Neanderthals, and human ancestors going
back 500,000 years. Several authors propose an alternative: population
replacement associated with cultural group selection. In this hypothesis,
total human population is always large. However effective population
size is small because a small fraction of groups expand, replace others,
and contribute most of the ancestry of later populations (Fig. 1.b).
Whitehead, Richerson, & Boyd, 2002) conduct a simulation of gene/
culture coevolution, and show that with (a) low rates of migration between groups, (b) substantial fitness variation associated with cultural
differences, and (c) low rates of cultural transmission between groups,
the “cultural hitchhiking” of genes associated with successful cultures
can produce low effective population sizes, in agreement with the genetic record. Premo and Hublin (2009) find similar results in a model
of culturally mediated migration, “the general mechanism whereby individuals can only migrate to groups that surpass a given level of cultural
familiarity” (p. 33). In their model, genes don't just hitchhike on cultural
expansion, but may drive it.
These authors don't specify the nature of the fitness advantage associated with different cultures. Archeologists, if they allow at all for differential cultural success and population replacement, commonly
favor materialist explanations, like environmental change or differences
in technology. For example, a popular explanation for population replacement in the Holocene involves the spread of agriculture
(Bellwood, 2005). Farming can support more people per square kilometer than foraging, and farmers tend to expand at the expense of foragers.
But in this case there are indications that the story is more complicated, and involves differences in social organization. The Austronesians, a classic case of a supposed farming expansion, now look more
like “an agricultural revolution that failed” (Blench, 2014). Early Austronesian speakers arriving in island Southeast Asia and Melanesia were
“fisher-foragers” more than they were farmers, pioneers lighting out
for the territory in order to reproduce a hierarchical sacred order. The
Bantu, another classic case of supposed First Farmers, apparently
showed up in East Africa later than earlier farmers and herders, with a
socio-cultural organization that facilitated the assimilation and replacement of earlier arrivals (Ehret, 1998).
More generally, major demic expansions seem to have spread distinctive social structures that did not merely piggyback on new technologies, but helped to propel the expansions (Jones, 2003, 2011). And in
historic times, the differential expansion of states and cultures seems
to have resulted not just from material advantages, but from differences
in the strength of ultrasocial norms (Turchin, Currie, Turner, & Gavrilets,
2013). (However in these cases population replacement typically affected elites more than commoners.)
4.2. The ethnocentric complex
The behavior genetic changes associated with a single episode of
demic expansion are likely to be modest, but as long as defection rates
are not too high, changes can accumulate over time. Eventually an evolutionary history of ethnic group selection should leave its stamp on
human psychology.
The psychology of ethnicity and ethnocentrism, according to the
model developed here, does not take the form of a uniform ethnicity
module or Darwinian algorithm. Instead, because no cultural regime is
reproduced perfectly and different regimes impose different selection
pressures, there is heritable variation in ethnocentrism. Also, the psychology of ethnocentrism, according to the model, is not simply an individual disposition to help fellow ethnics. It is a political psychology, a
disposition to maintain a particular social order – a group phenotype –
that regulates ethnic group cooperation, recruitment, and altruism.
Compare this with our emerging understanding of political psychology. On current evidence, individuals' political opinions are not just a
product of self-interested calculation or rational deliberation. Nor are
they just a historically contingent hodgepodge. Instead, opinions largely
reflect stable underlying individual variation in attitudes toward social
order, hierarchy, tradition, and in-groups and out-groups. These can
be ordered roughly along a right-left, conservative-liberal axis
(Duckitt, 1989; Hatemi & McDermott, 2012; Hibbing, Smith, & Alford,
2013; Tuschman, 2013). And the variation is heritable (Martin et al.,
1986; Smith, Oxley, Hibbing, Alford, & Hibbing, 2011).
For example, one line of research identifies a cluster of political
views that has been labeled Right Wing Authoritarianism (RWA)
(Altemeyer, 1996). RWA has three components, authoritarian submission, authoritarian aggression, and conventionalism. RWA is a stable individual disposition. And measures of RWA are valid across cultures,
although some of the associated policy particulars vary (de Regt,
Mortelmans, & Smits, 2011).
Related work by Haidt and coworkers (Graham, Haidt, & Nosek,
2009) points to five psychological systems underlying moral intuitions:
avoidance of harm, fairness, hierarchy, groupishness, and sacredness.
The first two principles are important both for liberals and conservatives, while the last three operate more strongly among conservatives.
Authoritarianism correlates with ethnocentrism (Altemeyer, 1996;
Kinder & Kam, 2009). Research on ethnicity consistently finds a general
dimension of prejudice: individuals with negative attitudes about one
outgroup usually have negative attitudes about other outgroups. (However, positive attitudes toward one's own group correlate imperfectly
with negative attitudes toward outgroups, both across individuals –
Kinder & Kam, 2009 – and across cultures – Cashdan, 2001.) Ethnocentric individuals show a higher degree of group identification in experimental settings (Perreault & Bourhis, 2016). Ethnocentrism and
stereotype endorsement also correlate with essentialist beliefs in the
discreteness, immutability, and biological basis of social categories
(Haslam, Bastian, Bain, & Kashima, 2006).
And ethnocentrism – measured as the difference between attitudes
toward one's own group and toward outgroups – predicts variation in
opinions on a number of issues (Kinder & Kam, 2009). For example, in
the United States, more ethnocentric whites are more likely to oppose
government programs that disproportionately benefit non-whites, but
to support broad based programs like Social Security.
In short, there is a heritable syndrome of political attitudes associated
with ingroup favoritism and the enforcement of social rules. More tentatively, this syndrome may include essentialist beliefs about social categories. Interestingly, conservative traits form a tighter phenotypic cluster
than liberal ones, suggesting that conservatism has been more of a target
of selection (Hibbing et al., 2013, pp. 223–224). All this is consistent with
the argument here, that ethnic group selection has shaped political psychology. More specifically, that an ethnocentrism syndrome has been
maintained because, in some societies at some times, individuals high
in ethnocentrism and conservatism have succeeded in setting up social
enforcement mechanisms that both favor their group in competition
with others and, concordantly (see Subsection 3.2), impose extra costs
on non-conforming, non-ethnocentric individuals within the group.
There are other possibilities, of course. It could be that genes “for” political attitudes were selected for something other than their political effects. Consider that political attitudes correlate with pre-political
personality traits like openness to experience (correlated with liberalism)
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
and conscientiousness (correlated with conservatism). Conceivably the
influence of genes on political attitudes might follow the sequence
genes → personality → politics, and the real evolutionary story might involve genes and personality, with political attitudes dragged along as a
byproduct. But the behavior genetic data show something else; they
show a direct connection running from genes to attitudes, independent
of personality (Hatemi & Verhulst, 2015; Verhulst, Eaves, & Hatemi,
2011). This suggests we need to think seriously about the coevolutionary dynamics of genes, political attitudes, and cultural regimes.
5. Conclusion
Both the study of prehistory and political psychology are changing
rapidly in the face of new evidence from biology, especially genetics. It
would be intellectually satisfying if we could integrate these findings
under the heading of an already existing theory, by equating ethnicity
with kinship and applying kin selection theory. But we've seen that
this won't work. Ethnicity, like kinship, may have to do with shared
genes, and there may be such things as ethnic genetic interests and ethnic nepotism. But an evolutionary theory of ethnicity – even the
barebones theory presented here – has to be something more than the
theory of kin selection, because of the way ethnicity is entangled with
some of the most complicated aspects of human sociality: norms,
rules, and political ideals, and the way they affect, and are affected by,
large-scale population processes.
Appendix A. Derivation of Condition (12), wi ¼ pi ð f −gÞ=SP
In a group in regime U with an initial frequency of P equal to p0, the
frequency of P will evolve over time according to a logistic equation:
p½t ' ¼
p0 ! esp !t
1−p0 þ p0 ! esp !t
ðA1Þ
The proportional growth rate for such a group is given by:
u0 ½t '
¼ sU −ð f ð1−p½t 'Þ þ g ! p½t 'Þ−z
u½t '
ðA2Þ
where z, which is the same for all U and V groups, is set to ensure that
the net change in in frequency across all groups sums to 0. Solving for
u[t] gives:
!
"ð f −gÞ=sp
u½t ' ¼ u0 ! eðsu − f −zÞt 1−p0 þ p0 ! esp !t
ðA3Þ
Over a long period of time, the quantity in parentheses is dominated
by the p0 ⋅ eSP ⋅t term, so:
u½t' ¼ u0 ⋅eðsU −g−zÞt ⋅p0 ðf −gÞ=sp ast→∞
(SU − g − z)t
ðA4Þ
We can normalize this by dividing by u0 ⋅ e
, the expected
long-term increase of a U group that starts out with u = u0 and p = 1,
to get:
ð f −gÞ=sp
wi ¼ pi
ðA5Þ
where wi is the normalized long-term reproductive contribution of a
group in regime U with frequency of allele P equal to pi. This is
Condition (12) in the text.
This derivation assumes a steady state distribution of U and V groups
with various pi, so that z is not a function of t. It ignores the indirect route
to cultural/genetic reproduction that occurs if defecting U groups, by
“seeding” regime V with P, increase the rate of counter-defection from
V to U.
9
References
Altemeyer, R. (1996). The authoritarian specter. Cambridge: Harvard University Press.
Ambrose, S. H. (1998). Late Pleistocene human population bottlenecks, volcanic winter,
and the differentiation of modern humans. Journal of Human Evolution, 34, 623–651.
Bell, A. V., Richerson, P. J., & McElreath, R. (2009). Culture rather than genes provides
greater scope for the evolution of large-scale human prosociality. Proceedings of the
National Academy of Science USA, 106, 17671–17674.
Bellwood, P. (2005). First farmers: The origins of agricultural societies. Oxford: Blackwell.
Blench, R. (2014). The Austronesians: An agricultural revolution that failed. International
conference on Formosan indigenous peoples: Contemporary perspectives (pp. 15).
Bowles, S. (2006). Group competition, reproductive leveling, and the evolution of human
altruism. Science, 314, 1569–1572.
Brigant, I. (2001). The homeopathy of kin selection: An evaluation of van den Berghe's
sociobiological approach to ethnic nepotism. Politics and the Life Sciences, 20,
203–215.
Cashdan, E. (2001). Ethnocentrism and xenophobia: A cross-cultural study. Current
Anthropology, 42, 760–764.
Connor, W. (1993). Ethnonationalism: The quest for understanding. Princeton: Princeton
University Press.
Crispo, E. (2007). The Baldwin effect and genetic assimilation: Revisiting two mechanisms
of evolutionary change mediated by phenotypic plasticity. Evolution, 61, 2469–2479.
Dawkins, R. (1979). Twelve misunderstandings of kin selection. Zeitschrift für
Tierpsychologie, 51, 184–200.
Duckitt, J. (1989). Authoritarianism and group identification: A new view of an old construct. Political Psychology, 10, 63–84.
Ehrenreich, I. M., & Pfennig, D. W. (2015). Genetic assimilation: A review of its potential
proximate causes and evolutionary consequences. Annals of Botany(Special Issue), 1–11.
Ehret, C. (1998). An African classical age: Eastern and Southern Africa in world history
1000 BC to AD 400. University of Virginia Press.
Falconer, D. S., & McKay, T. (1996). Introduction to quantitative genetics (4th Edition ).
London: Pearson.
Frank, S. (1998). Foundations of social evolution. Princeton: Princeton University Press.
Graham, J., Haidt, J., & Nosek, B. A. (2009). Liberals and conservatives rely on different sets
of moral foundations. Journal of Personality and Social Psychology, 96, 1029–1046.
Hamilton, W. D. (1964). The genetical evolution of social behavior, I. Journal of Theoretical
Biology, 7, 1–16.
Harpending, H. (2002). Kinship and population subdivision. Population and Environment,
24, 141–147.
Harpending, H., Batzer, M., Gurven, M., Jorde, L., Rogers, A., & Sherry, S. T. (1998). Genetic
traces of ancient demography. Proceedings of the National Academy of Sciences USA, 95,
1961–1967.
Hartshorn, M., Kaznatcheev, A., & Schultz, T. (2013). The evolutionary dominance of ethnocentric cooperation. Journal of Artificial Societies and Social Simulation, 16, 7.
Haslam, N., Bastian, B., Bain, P., & Kashima, Y. (2006). Psychological essentailism, implicit
theories, and intergroup relations. Group Processes & Intergroup Relations, 9, 63–76.
Hatemi, P., & Verhulst, B. (2015). Political attitudes develop independently of personality
traits. PloS One, 10, e0118106. https://doi.org/10.1371/journal.pone.0118106.
Hatemi, P. K., & McDermott, R. (2012). The genetics of politics: Discovery, challenges,
progress. Trends in Genetics, 28, 525–533.
Hibbing, J. R., Smith, K., & Alford, J. R. (2013). Predisposed: Liberals, conservatives, and the
biology of political differences. Routledge.
Horowitz, D. L. (1985). Ethnic groups in conflict. Berkeley: University of California Press.
Jones, D. (2000). Group nepotism and human kinship. Current Anthropology, 41, 303–319.
Jones, D. (2003). Kinship and deep history: Exploring connections between genes, culture
areas, and languages. American Anthropologist, 105, 501–514.
Jones, D. (2011). The matrilocal tribe: An organization of demic expansion. Human Nature,
22, 177–200.
Jones, D. (2016). Socially enforced nepotism: How norms and reputation can amplify kin
altruism. PloS One, 11, e0155596. https://doi.org/10.1371/journal.pone/0155596.
Kinder, D. R., & Kam, C. D. (2009). Us against them: Ethnocentric foundations of American
opinion. Chicago: University of Chicago Press.
Leimar, O., & Hammerstein, P. (2001). Evolution of cooperation through indirect reciprocity. Proceedings of the Royal Society B, 268, 745–753.
Lewontin, R. C. (1972). The apportionment of human diversity. Evolutionary Biology, 6,
381–398.
Martin, N. G., Eaves, L. J., Heath, A. C., Jardine, R., Feingold, L. M., & Eysenck, H. J. (1986).
Transmission of social attitudes. Proceedings of the National Academy of Sciences
USA, 83, 4364–4368.
McElreath, R., Boyd, R., & Richerson, P. J. (2003). Shared norms and the evolution of ethnic
markers. Current Anthropology, 44, 122–130.
Norenzayan, A. (2015). Big gods: How religion transformed cooperation and conflict.
Princeton: Princeton University Press.
Nowak, M., & Sigmund, K. (1998). Evolution of indirect reciprocity by image scoring.
Nature, 393, 573–577.
Perreault, S., & Bourhis, R. Y. (2016). Ethnocentrism, social identification, and discrimination. Personality and Social Psychology Bulletin, 25, 92–103.
Premo, L. S., & Hublin, J. -J. (2009). Culture, population structure, and low genetic diversity
in Pleistocene hominins. Proceedings of the National Academy of Sciences USA, 106, 33–37.
de Regt, S., Mortelmans, D., & Smits, T. (2011). Left-wing authoritarianism is not a myth,
but a worrisome reality: Evidence from 13 Eastern European countries. Communist
and Post-Communist Studies, 44, 299–308.
Richerson, P., Baldini, R., Bell, A. V., Demps, K., Frost, K., Hillis, V., ... Zefferman, M. (2016).
Cultural group selection plays an essential role in explaining human cooperation: A
sketch of the evidence. Behavioral and Brain Sciences, 39, e30. https://doi.org/10.
1017/S0140525X1400106X.
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
10
D. Jones / Evolution and Human Behavior xxx (2017) xxx–xxx
Rogers, A. (1990). Group selection by selective emigration: The effects of migration and
kin structure. The American Naturalist, 135, 398–413.
Rushton, J. P. (2005). Ethnic nationalism, evolutionary psychology and Genetic Similarity
Theory. Nations and Nationalism, 4, 489–507.
Salter, F., & Harpending, H. (2013). J. P. Rushton's Theory of ethnic nepotism. Personality
and Individual Differences, 55, 256–260.
Shaw, R. P., & Wong, Y. (1989). Genetic seeds of warfare: Evolution, nationalism, and patriotism. London: Unwin Hyman.
Smith, K. B., Oxley, D. R., Hibbing, M. V., Alford, J. R., & Hibbing, J. R. (2011). Linking genetics and political attitudes: Reconceptualizing political ideology. Political Psychology,
32, 369–397.
Sturm, R. A. (2009). Molecular genetics of human pigmentation diversity. Human
Molecular Genetics, 18, R9–R17.
Turchin, P., Currie, C. E., Turner, E. A. L., & Gavrilets, S. (2013). War, space, and the evolution of old world complex societies. Proceedings of the National Academy of Science
USA, 110, 16384–16389.
Tuschman, A. (2013). Our political nature: The evolutionary origins of what divides us. Amherst, New York: Prometheus Books.
Van Den Berghe, P. (1981). The ethnic phenomenon. New York: Elsevier.
Vanhanen, T. (1999). Ethnic conflicts explained by ethnic nepotism. Bingley, UK: Emerald
Group Publishing.
Verhulst, B., Eaves, L. J., & Hatemi, P. K. (2011). Correlation not causation: The relationship
between personality traits and political ideologies. American Journal of Political Science., 56, 34–51.
Weber, M. (1978). Economy and society: An outline of interpretive sociology. vol. 1.Berkeley:
University of California Press.
Whitehead, H., Richerson, P., & Boyd, R. (2002). Cultural selection and genetic diversity in
humans. Selection, 3, 115–125.
Whitmeyer, J. M. (1997). Endogamy as a basis for ethnic behavior. Sociological Theory, 15,
162–178.
Please cite this article as: Jones, D., Kin selection and ethnic group selection, Evolution and Human Behavior (2017), https://doi.org/10.1016/
j.evolhumbehav.2017.08.004
View publication stats