Sex equality can explain the
unique social structure of
M. Dyble,*G. D. Salali, N. Chaudhary, A. Page, D. Smith, J. Thompson, L. Vinicius,
R. Mace, A. B. Migliano
The social organization of mobile hunter-gatherers has several derived features,
including low within-camp relatedness and fluid meta-groups. Although these features
have been proposed to have provided the selective context for the evolution of human
hypercooperation and cumulative culture, how such a distinctive social system may have
emerged remains unclear. We present an agent-based model suggesting that, even if all
individuals in a community seek to live with as many kin as possible, within-camp
relatedness is reduced if men and women have equal influence in selecting camp members.
Our model closely approximates observed patterns of co-residence among Agta and
Mbendjele BaYaka hunter-gatherers. Our results suggest that pair-bonding and increased
sex egalitarianism in human evolutionary history may have had a transformative effect
on human social organization.
Contemporary mobile hunter-gatherers co-
operate extensively with unrelated individ-
uals across multiple social and economic
domains. Many communities of mobile
hunter-gatherers (hereafter hunter-gatherers)
share food extensively within camp and hunt,
gather, and fish cooperatively (1). Alloparenting
is also commonplace (2,3).Theimportanceof
cooperative activities is reflected in many hunter-
gatherer societies as a pervasive ethic of egalitarian-
ism (4,5). Like a number of non-human primate
species, humans live in multimale, multifemale
groups (6). However, we maintain enduring pair
bonds, resulting in what have been described as
“multifamily”groups (7). In addition, and in con-
trast to the bounded and territorial groups of
chimpanzees (8,9), bonobos (10) andgorillas (11),
contemporary hunter-gatherers have fluid social
networks where family units are relatively auton-
omous, with couples and their children moving
often between bands (12), living with kin of
either the husband or the wife. This residence
pattern has been described as either “bilocal”or
As well as being highly mobile, contemporary
hunter-gatherer camps include a significant pro-
portion of unrelated individuals (14) and are less
closely related than groups of non-foraging small-
scale societies (15). Given the inclusive fitness
benefits of living with kin, why hunter-gatherers
live with unrelated individuals is a puzzle, even
more so if one considers that hunter-gatherers
show a preference for living with siblings (13)and
preferentially include kin in their campmate choices
and social networks (16). Therefore, the mecha-
nisms by which contemporary hunter-gatherers
attempt to maximize co-residence and cooperation
with kin, but nonetheless end up residing mostly
with unrelated individuals, remain unclear.
Here, we offer a solution for this apparent par-
adox by demonstrating that, even where all indi-
viduals are actively assorting with kin, within-group
relatedness is reduced if both sexes have influ-
ence over camp composition, as is the case among
egalitarian, multilocal hunter-gatherers. We present
a simulation of camp assortment where individ-
uals attempt to reside with as many kin as pos-
sible under two conditions. In the egalitarian
condition, men and women have equal influ-
ence on camp composition, whereas in the non-
egalitarian condition only one sex has influence.
We compared the results with previously unpub-
lished data from two hunter-gatherer groups, the
Palanan Agta (N= 4055 dyads) and Mbendjele
ing population, the Paranan (N=1049dyads).
We demonstrate that low within-camp relatedness
emerges naturally from men and women seeking
to maximize the presence of related kin. In con-
trast, in societies where decision-making on co-
residence rests on one sex only, as in the case
of patrilocal farmers, low relatedness does not
emerge. Our model offers a mechanism that re-
conciles individual-level preferences for kin with
reduced camp-level relatedness. Assuming that
extant hunter-gatherers live in social structures
resembling the ones existing in past hominins,
our model explains how the shift from an an-
cestral hierarchical, female-dispersal system, to
a multilocal, egalitarian one would provide the
selective context for expanded social networks,
cumulative culture, and cooperation among un-
SCIENCE sciencemag.org 00 MONTH 2015 •VOL 000 ISSUE 0000 1
University College London (UCL) Anthropology, 14 Taviton
Street, London WC1H 0BW, UK.
*Corresponding author. E-mail: firstname.lastname@example.org
Fig. 1. Co-residence patterns across modeled and observed egalitarian populations. Chart area represents the proportion of all dyads across nine
categories of relatedness for the egalitarian model (left), Agta (middle left), Mbendjele (middle right), Ache (bottom right), and Ju/’hoansi (top right). Ache
and Ju/’hoansi data redrawn from Hill et al. (2011).
MS no: REaaa5139/CF/ANTHRO
Among the Agta, we collected data from 191
adults across 11 camps, coding a total of 4055
dyadic relationships. Among the Mbendjele, we
collected data from 103 adults across nine camps,
totaling 1863 dyadic relationships. Mean experi-
enced camp size was 18.09 adults (SD = 8.62) for
the Mbendjele and 21.23 adults (SD = 8.61) for
the Agta. Both populations were multilocal, with
husbands and wives living with similar numbers
of consanguineal (genetic) kin (table S1 and fig.
S1). In both groups, around 25% of dyads repre-
sented consanguineal kin, 25% were close affinal
kin, and around 50% of dyads were distant af-
final kin or unrelated individuals (
F1 Fig. 1 and
table S2). These results are similar to those re-
ported for the Ache and Ju/’hoansi by Hill and
colleagues (14); see Fig. 1.
In contrast to the unbiased residence patterns
of the Agta and Mbendjele, Paranan farmers (n=
49 adults, 1049 dyads) demonstrate a significant
male bias in residence, with men living with a
larger number of primary kin (n= 23, mean =
2.65, SD = 2.29) than women (n=26,mean=
1.27, SD = 2.05; P=0.031).Despitehavingacom-
parable group size of 21.4 adults (SD = 9.30), the
Paranan live with fewer unrelated individuals
than the hunter-gatherers (4.2% versus 16.7%)
=108.93,P< 0.001) (F2 Fig. 2).
Although it is possible that low within-camp
relatedness could result from random dispersal,
with households moving randomly between camps
and living with related individuals only by chance,
our results do not suggest that this is the case.
Rather, the observed frequency of primary kin
co-residence was significantly higher than would
be expected if individuals assorted randomly
across camps (Mbendjele, c
= 982.00, P< 0.001). Thus, hunter-
gatherer co-residence patterns are notable not
only in their low-relatedness but because this
low relatedness occurs despite the positive assort-
ment of kin.
We developed a model to understand how
hunter-gatherers come to co-reside with a large
number of unrelated individuals at the group
level, despite a preference toward living with kin
at the individual level. We ran two versions of the
model: one egalitarian, where both husband and
wife have equal influence over where their house-
hold resides, and a non-egalitarian one, where
only one sex has influence. Even at relatively
small population sizes, these two conditions re-
sult in large differences in group composition.
Across 100 simulations at a population size of 20,
for example, there was a significantly larger
proportion of unrelated dyads in the modeled
egalitarian camps (12.0% T8.4) compared with
the non-egalitarian, single-sex dispersal camps
(0.6% T1.5) (c
= 4372.36, P<0.001;Figs.1and2).
Although it is known that group relatedness de-
creases with increased group size (15), modeled
egalitarian camps show higher proportions of
unrelated individuals irrespective of camp size
F3 Fig. 3).
The modeled co-residence patterns also mirror
our observed data. The proportion of unrelated
dyads in the model at a comparable group size
(n=20agents)(12.0%T8.4) was not significantly
different from the observed proportion of unre-
lated co-residency among the Agta (11.2%, c
1.98, P= 0.016). Although the Mbendjele had sig-
nificantly larger numbers of unrelated individuals
= 440.76, P< 0.001), this was in the direction
consistent with our hypothesis. The obser ved
proportion of unrelated dyads among the Paranan
(4.2%) was larger than the modeled proportion
(0.6% T1.5, c
than either of the observed hunter-gatherer pop-
ulations (see above) and the egalitarian model
= 58.65, P<0.001).
Our results suggest that pair-bonding alone is
not sufficient to explain the low levels of related-
ness seen in hunter-gatherer groups. Rather, both
pair-bonding and sex equality in residential decision-
making act together to constrain the overall re-
latedness of groups, leading to the co-residence
of individuals unrelated through either genetic
or affinal ties.
It has been proposed elsewhere that the com-
bination of pair-bonding, cooperation among un-
related males, and increased mobility derived from
male alliances could account for the low related-
ness of hunter-gatherer camps (7). We argue instead
that low within-camp relatedness is a consequence
of sex equality in hunter-gatherer couples, with
husbands and wives having an equal influence
over camp composition. Given sex equality, we
have shown that unrelated individuals come to
200 MONTH 2015 •VOL 000 ISSUE 0000 sciencemag.org SCIENCE
Fig. 2. Co-residence patterns across modeled and observed non-egalitarian populations. Chart
area represents the proportion of all dyads across nine categories of relatedness for the non-egalitarian
model (left) and Paranan (right).
Fig. 3. Modeling re-
latedness and equal-
ity. Results of the
egalitarian model (top)
camp sizes between
10 and 80. From
bottom to top, areas
(genetic) kin, affinal kin,
distant affinal kin, and
Exact proportions are
given in tables S3
10 20 30 40 50 60 70
10 20 30 40 50 60 70 80
MS no: REaaa5139/CF/ANTHRO
co-reside even when they display a strong indi-
vidual preference to live with kin, exemplified in
hunter-gatherers by the frequent co-residence of
brothers and sisters (14)andthehigherfrequen-
cy of related individuals in campmate and gift
networks (16). Therefore, our simulations provide
a mechanism for the emergence of low within-
the apparent contradiction between individual-
level preferences for living with kin and group-
level co-residence with non-kin. Gender inequality
reappeared in humans with the transition to
agriculture and pastoralism (17). Once heritable
resources, such as land and livestock, became
important determinants of reproductive success,
sex-biased inheritance and lineal systems started
to arise, leading to wealth and sex inequalities
(18). This predicted effect was demonstrated in
our non-egalitarian model and data from Paranan
agriculturalists. Our results also provide further
evidence that multilocality, rather than patrilocal-
ity, is the norm among mobile hunter-gatherers.
Understanding hunter-gatherer sex egalitari-
anism and the shift from hierarchical male philo-
patry typical of chimpanzees and bonobos to a
multilocal residence pattern is key to theories of
human social evolution. A possible clue for the
evolution of sex equality in the hominin lineage
was the increase in the cost of human reproduc-
tion associated with larger brain sizes in early
Homo (19). Higher offspring costs would require
investment from both mothers and fathers (20),
as seen among extant hunter-gatherers (3,21).
The need for biparental investment predicts in-
creased sex equality (22), reflected in the high
frequency of monogamy and the reproductive
schedules of male hunter-gatherers, who typi-
cally stop reproducing early and exhibit long
life spans after their last reproduction, in con-
trast to male farmers and pastoralists, whose
reproductive spans extend well into late life (23).
The recognition of affinal ties throughout our
long life span has been argued to be an impor-
tant step in human social evolution, and house-
hold residence may also be influenced by a tug of
war between a husband and his affinal kin, who
The possibility of recruiting help from both ma-
ternal and paternal kin by moving camps might
have been an important adaptation to meet
reproductive costs in unpredictable environments,
for example, by increasing the frequency of co-
residence with grandmothers, who have an im-
portant provisioning role in many hunter-gatherer
societies (24). Increased reproductive costs, coop-
erative breeding, and sex equality in residential
decision-making can explain why hunter-gatherer
parents live in groups containing multiple mated
pairs, why hunter-gatherers recruit help both from
related and unrelated individuals, and why hunter-
gatherer camps exhibit low levels of relatedness.
Sex equality and the resulting low within-camp
relatedness had many important consequences.
Co-residence with unrelated individuals set the
selective environment for the evolution of hyper-
cooperation and prosociality (25). Sex equality
suggests a scenario where cooperation among
unrelated individuals can evolve in the absence
of wealth accumulation, reproductive inequalities,
and intergroup warfare (26). Couples freely mov-
ing between camps and sharing interests with
kin and affines would be able to maintain coop-
eration without the need for more complex sys-
tems, such as cultural group selection and altruistic
Last, this social system may have allowed
hunter-gatherers to extend their social networks,
buffering environmental risk and promoting lev-
els of information exchange required for cumu-
lative culture (28–31).
REFERENCES AND NOTES
1. M. Gurven, Behav. Brain Sci. 27, 543–583 (2004).
2. P. K. Ivey, Curr. Anthropol. 41, 856–866 (2000).
3. C. L. Meehan, R. Quinlan, C. D. Malcom, Am. J. Hum. Biol. 25,
4. J. Woodburn, Man 17, 431–451 (1982).
5. J. Lewis, in Hunter-Gatherers of the Congo Basin: Cultures,
Histories, and Biology of African Pygmies, B. S. Hewlett, Ed.
(Transaction, City, NJ, 2014), pp. 219–244.
6. L. Rodseth, R. W. Wrangham, A. M. Harrigan, B. B. Smuts,
Curr. Anthropol. 32, 221–254 (1991).
7. B. Chapais, Primeval Kinship: How Pair-Bonding Gave Birth to
Human Society (Harvard Univ. Press, Cambridge, MA, 2009).
8. M. L. Wilson, R. W. Wrangham, Annu. Rev. Anthropol. 32,
9. J. Mitani, D. Watts, Behaviour 138, 299–327 (2001).
10. R. W. Wrangham, D. Peterson, Demonic Males: Apes and the
Origins of Human Violence (Houghton Mifflin Harcourt, City, 1996).
11. P. Sicotte, Am. J. Primatol. 30,21–36 (1993).
12. R. Kelly, J. Anthropol. Res. 39, 277–306 (1983).
13. F. Marlowe, Curr. Anthropol. 45, 277–284 (2004).
14. K. R. Hill et al., Science 331, 1286–1289 (2011).
15. R. S. Walker, Evol. Hum. Behav. 35, 384–388 (2014).
16. C. L. Apicella, F. W. Marlowe, J. H. Fowler, N. A. Christakis,
Nature 481, 497–501 (2012).
17. M. Martin, B. Voorhies, Female of the Species (Columbia Univ.
Press, New York, 1975).
18. R. Mace, Evol. Anthropol. 22, 251–258 (2013).
19. L. Aiello, P. Wheeler, Curr. Anthropol. 36, 199–221 (1995).
20. H. Kaplan, K. Hill, J. Lancaster, A. M. Hurtado, Evol. Anthropol.
9, 156–185 (2000).
21. K. Hill, A. M. Hurtado, Proc. Biol. Sci. 276, 3863–3870
22. H. S. Kaplan, J. B. Lancaster, in Offspring: Human Fertility
Behavior in Biodemographic Perspective, K. W. Wachter,
R. A. Bulatao, Eds. (National Academies Press, Washington,
DC, 2003), pp. 170–223.
23. L. Vinicius, R. Mace, A. Migliano, PLOS ONE 9, e112236 (2014).
24. K. Hawkes, J. F. O’Connell, N. G. Jones, H. Alvarez,
E. L. Charnov, Proc. Natl. Acad. Sci. U.S.A. 95,1336–1339
25. J. M. Burkart et al., Nat. Commun. 5, 4747 (2014).
26. D. P. Fry, P. Söderberg, Science 341, 270–273 (2013).
27. H. M. Lewis, L. Vinicius, J. Strods, R. Mace, A. B. Migliano,
Nat. Commun. 5, 5789 (2014).
28. M. Kempe, A. Mesoudi, Evol. Hum. Behav. 35, 285–290
29. K. R. Hill, B. M. Wood, J. Baggio, A. M. Hurtado, R. T. Boyd,
PLOS ONE 9, e102806 (2014).
30. R. Lee, The! Kung San (Cambridge Univ. Press, Cambridge,
31. R. Tonkinson, The Mardu Aborigines (Holt, Rinehard, and
Winston, City, 1978).
ACKNO WLED GME NTS
This project was funded by the Leverhulme Trust grant
RP2011-R-045 to A.B.M. R.M. also received funding from European
Research Council Advanced Grant AdG 249347. A.B.M.
conceived the project; M.D. created the model; M.D., A.P.,
G.D.S., N.C., A.B.M., D.S., and J.T. collected the data; M.D. and
G.D.S. wrote scripts and analyzed the data. M.D., A.B.M, L.V.,
and R.M. wrote the paper. We thank P. Gerbault, J. Stevenson,
J. Lewis, and R. Schlaepfer for help in the field and the Human
Evolutionary Ecology Group at UCL as well as three anonymous
reviewers for valuable comments on the manuscript. Last, we
thank our assistants in both Congo and the Philippines as well
as the Agta, Paranan, and Mbendjele communities.
Materials and Methods
Figs. S1 and S2
Tables S1 to S9
18 December 2014; accepted 16 April 2015
SCIENCE sciencemag.org 00 MONTH 2015 •VOL 000 ISSUE 0000 3
MS no: REaaa5139/CF/ANTHRO