The architecture of human kin detection
, John Tooby
& Leda Cosmides
Evolved mechanisms for assessing genetic relatedness have been found in many species, but their existence in humans has
been a matter of controversy. Here we report three converging lines of evidence, drawn from siblings, that support the
hypothesis that kin detection mechanisms exist in humans. These operate by computing, for each familiar individual, a
unitary regulatory variable (the kinship index) that corresponds to a pairwise estimate of genetic relatedness between self
and other. The cues that the system uses were identified by quantitatively matching individual exposure to potential cues of
relatedness to variation in three outputs relevant to the system’s evolved functions: sibling altruism, aversion to personally
engaging in sibling incest, and moral opposition to third party sibling incest. As predicted, the kin detection system uses two
distinct, ancestrally valid cues to compute relatedness: the familiar other’s perinatal association with the individual’s
biological mother, and duration of sibling coresidence.
For the past 50 years, evolutionary biologists have argued that
genetic relatedness should have played a role in the social evolution
of species, such as humans, in which close genetic relatives frequently
. According to kin selection theory, computational variants
that allocate altruistic effort effectively with respect to kinship out-
compete variants that fail to regulate behaviour conditionally in
response to relatedness. The effects of relatedness have been docu-
mented in a great diversity of taxa, ranging from social amoebas
, to birds
. To regulate behaviour conditionally in response to
different degrees of kinship, organisms require mechanisms to dis-
criminate genetic relatedness. Such mechanisms have been discov-
ered in a variety of nonhuman species
Equally, in long-lived, low-fecundity species with an open breed-
ing structure (such as humans), the fitness of offspring is strongly
affected by how closely parents are related. In such species, concept-
ive sexual behaviour between close genetic relatives produces off-
spring that suffer from inbreeding depression—a decline in fitness
caused by rendering more deleterious recessives homozygous
and aggravated by parasites targeting more genetically homogeneous
sets of hosts
. Consequently, heritable variants that cost-effectively
reduce inbreeding depression by avoiding mating with close genetic
relatives outcompete variants in which mating decisions are un-
affected by relatedness.
The socioecology and population biology of human foragers
suggest that our ancestors would have been subject both to inbreed-
ing depression and kin selection. This leads to the prediction that
humans have an evolved system for detecting genetic relatedness,
coupled to two output systems: one regulating altruism, the other
regulating mate choice. Yet, there has been little research into the
existence and design of human kin detection mechanisms
The best-known exceptions are a handful of anthropological stud-
ies testing Westermarck’s prescient 1891 hypothesis
exposure during childhood weakens sexual attraction among adults.
These documented that non-relatives raised together in exceptional
developmental circumstances (for example, cre
`che-mates or children
cohabiting with future spouses) show lower rates of marriage or
marital fertility, and higher rates of divorce and infidelity—archivally
derived sociological measures used as proxies for the intensity of
. But to map the information-processing architecture
of a system predicted to detect genetic relatedness—and see whether
it regulates altruistic as well as sexual motivation—it is necessary to
measure the responses of living individuals drawn from a more
species-characteristic range of family compositions, such as those
that include actual genetic relatives.
Accordingly, the goal of the studies reported here was to test for the
existence of a human kin detection system, and to test a series of basic
predictions about its design features and architecture. It is ethically
unacceptable to subject humans to the life-changing experimental
manipulations used to discover kin detection systems in other spe-
cies. So the architecture was mapped by quantitatively matching
individual variation in the two predicted output systems—sibling
altruism and opposition to incest—to naturally generated individual
variation in developmental parameters that were predicted to serve as
cues of relatedness.
Model of architecture and predictions
We propose that, for each familiar individual, i, the kin detection
system computes and updates a continuous variable, the kinship
, that corresponds to the system’s pairwise estimate of
genetic relatedness between self and i. These computational elements
are regulatory variables that serve as input to neural programs regu-
lating altruism towards iand, separately, to programs regulating
sexual behaviour towards i.
Because relatedness cannot be directly observed, the system must
be designed to register cues relevant to determining relatedness. To
compute the kinship index, the system requires (1) monitoring cir-
cuitry designed to register cues to relatedness, and (2) a computa-
tional device, the kinship estimator, whose procedures have been
tuned by a history of selection to take these registered inputs and
transform them into a kinship index.
The cues the system uses cannot simply be derived ontogenetically
(‘learned’) by identifying which arbitrary and transient cues happen
to best predict relatedness in the local environment. To do this, the
system would have to already know the relatedness of others—the
very problem it needs to solve. Instead, the kin detection system must
contain within its evolved design a specification of the core cues that
it will use to determine relatedness—cues that reliably tracked genetic
relatedness in the ancestral social environments that selected for the
kin detection system.
Center for Evolutionary Psychology, University of California Santa Barbara, Santa Barbara, California 93106, USA.
Department of Psychology, University of Hawaii, Honolulu, Hawaii
15 February 2007
©2007 Publishing Group
For human foragers, a potentially informative cue to kinship is
provided by the close perinatal association between mother and neo-
nate that begins with birth and is enforced by the exigencies of early
mammalian maternal care. Maternal perinatal association (MPA)
provides a basis for the reliable mutual detection of mother and
offspring and can, in turn, be used as an anchor point for sibling
detection. Ancestrally, if an individual observed an infant in a dur-
able, perinatal association with the individual’s mother, then it was
highly probable that that infant was the individual’s sibling. We
therefore proposed that sibling detection includes a monitoring sub-
system specialized for registering MPA.
Although MPA is likely to be the single most informative cue, it
cannot be used (for example) by younger siblings, because they are
not alive at the time their older siblings are born and nursed. When
MPA is unavailable, the kinship estimator should fall back on other
cues that were highly predictive ancestrally. We predicted that the kin
detection system would include a second subsystem specialized for
registering the cumulative duration of coresidence summed over the
full period they receive parental care. Ancestrally, parents (especially
mothers) maintained close association with their children to care for
them, and for this reason siblings co-associate statistically more than
non-siblings. (Indeed, given the fusion–fission pattern of hunter–
gatherer association, this same variable should—to some extent—
link progressively more distant genetic relatives to increasingly
diluted motivational residues.) Among human foragers, the main-
tenance of parental proximity for care delivery begins with birth and
tapers off in late adolescence, a time when offspring become nearly
independent adult foragers and when mating motivates new patterns
. Although this hypothesis differs from the etho-
logical proposal of a period of early childhood imprinting
consistent with evidence that suggests that familiarity is a cue medi-
ating kin detection in non-human primates
The kinship estimator consists of algorithms for transforming the
registered cues into the kinship index, a variable whose magnitude
tracks relatedness between self and other. If the cues are integrated
into a single index, then we should find that the same patterns of
inputs are associated with the same patterns of outputs for both
altruism and sexual aversion. This model (summarized in Fig. 1)
leads to the following predictions.
(1) When MPA is absent, coresidence duration before adulthood
with an individual should (a) upregulate altruism towards that indi-
vidual, (b) upregulate sexual aversion towards that individual, and,
as a by-product, (c) upregulate moral opposition
(2) When MPA is present, it should produce the same three effects.
Selection should have tuned the procedures in the kinship estim-
ator to use MPA and coresidence in a way that takes account of
their relative informativeness and availability. Because MPA is the
more robust, higher quality cue, we expect that when both are
available, coresidence will be weighted by the kinship estimator far
less than MPA, and perhaps not at all. Therefore, we propose a third
(3) When MPA is present, coresidence duration will not be as
strong a predictor of altruistic motivations and sexual aversions.
That is, the kinship estimator will use MPA in preference to coresi-
dence duration in computing kinship.
Multiple, converging tests involving over 600 subjects were employed
to assess whether particular developmental parameters (including
MPA and coresidence duration) serve as cues to kinship and regulate
both kin-directed altruism and sexual avoidance. Participants
responded to questions regarding family composition and sibling
interactions and were asked to complete instruments measuring:
(1) frequency of altruistic behaviours towards a given sibling; (2)
the intensity of altruistic motivation towards a given sibling; (3)
the level of disgust evoked by the prospect of engaging in sexual acts
with a given sibling, and (4) how morally wrong they perceive sibling
incest among third parties to be (an unobtrusive measure of sexual
aversion towards siblings
The most important findings are displayed in Figs 2 and 3, which
show that each of the two predicted cues of genetic relatedness for
siblings—coresidence duration and maternal perinatal association—
regulate outputs from the two functionally independent motiva-
tional systems (altruism and incest aversion) in the predicted way.
(see Supplementary Information section 1).
Additional cue monitoring circuitry
(major histocompatibility complex?
Proposed model of the computational architecture of sibling
detection. Cues to kinship are registered by cue monitoring circuits, which
deliver their outputs to a kinship estimator. The kinship estimator uses these
cues to compute the magnitude of a regulatory variable—a kinship index—
for each individual, i, who is a potential sibling. The kinship index feeds into
programs that regulate sibling altruism and sexual aversion.
Effect size: coresidence duration
(for example, detection of
(for example, detection of
Moral opposition to incest
Sexual disgust (rank)
Sexual disgust (Likert; men)
Converging evidence indicates that the same computational
variable, the kinship index, regulates disparate kin-relevant behaviours.
The x-axis divides subjects into two groups—those who observed their
mothers caring for their sibling as a neonate (MPA cue present) and those
who did not (MPA cue absent). The y-axis shows the size of the correlation
between coresidence duration and each dependent measure. Duration of
coresidence predicts, with similar effect sizes, altruism and sexual aversions
only when the cue of maternal perinatal association (MPA) is absent, as it is
when younger siblings are detecting older ones. When the MPA cue is
present, coresidence duration fails to predict sibling directed behaviours.
This pattern appears for all measures: behavioural altruism, dispositional
altruism, sexual disgust and moral judgments of sibling incest. Adaptive
regulation of two distinct motivational output systems by the same pattern
of inputs implicates a common underlying regulatory variable (see also
Supplementary Information section 7).
15 February 2007
©2007 Publishing Group
The overall pattern of results was the same for men and women.
For this reason, results are reported for both sexes combined, unless
otherwise specified (see Methods).
When MPA is absent. When the MPA cue is absent—as is true
whenever youngers are detecting older siblings—coresidence dura-
tion significantly predicts altruistic motivations and, separately,
opposition to first and third person incest (Fig. 2). Subject’s duration
of coresidence with a particular sibling was positively correlated
with all outcome measures: how much the subject helps that sibling
(altruism: behavioural, P56310
(or 8 310
, see Methods and
Supplementary Information section 9), N5185; dispositional,
), N5185); how disgusted the subject is at
imagining sexual contact with that (opposite sex) sibling (sexual
disgust (rank), P50.0002 (0.0003), N5114; sexual disgust
(Likert; men), P50.0007 (0.0009), N5156, see Methods); and
how morally wrong the subject judges third party sibling incest
(moral opposition to incest, P50.003 (0.004), N547; see also refs
28, 29). Figure 2 shows that the effect sizes (r) for coresidence are very
similar across widely divergent outcome variables, as would be
expected if separate systems for altruism and sexual aversion were
being regulated by the same internal variable, a kinship index.
When MPA is present. When the MPA cue is present—which can
only be true for olders detecting younger siblings—levels of altruism
and sexual aversion are high (Supplementary Information section 1).
But in the presence of MPA, coresidence duration no longer predicts
a single outcome measure (effect sizes ,0; Fig. 2; Supplementary
Information section 2). Directed univariate analyses show that
MPA and coresidence interact (sexual disgust (Likert; men),
P50.02; sexual disgust (rank), P50.003; altruism (see Methods),
P50.03; moral opposition, P50.12; see Supplementary Infor-
mation section 1); the dramatic drop in effect sizes (all significant;
Supplementary Information section 2) seen in Fig. 2 demonstrates
that coresidence duration robustly affects altruism and sexual aver-
sion in the absence, but not in the presence, of the MPA cue.
MPA versus coresidence. MPA can only be observed by older sib-
lings, and so they are the only individuals who can potentially be
exposed to both MPA and coresidence duration cues. Thus analysis
of olders allows one to see how the kinship estimator integrates these
two cues to genetic relatedness.
Because MPA (as operationalized on these tests) is a dichotomous
variable (1, 0) with 84% of older siblings scoring 1, its effects are most
sensitively detected by using those outcome variables that are con-
tinuous and with high variance: altruism and moral opposition.
The study assessing altruism yielded the most subject-and-younger
sibling pairs (N5128). As Fig. 3 shows, MPA significantly pre-
dicted altruism towards younger siblings (r50.32, P50.0001
(0.00013)), even when controlling for coresidence (partial r50.22,
P50.006 (0.008), tolerance, 0.56, that is, much greater than the 0.10
collinearity threshold). This is important, because MPA and coresi-
dence duration are themselves correlated (r50.66). In contrast, the
relationship between coresidence duration and altruism towards
younger siblings (r50.24) disappears when the effects of MPA are
partialled out (partial r50.04, P50.33 (0.41)). When MPA, co-
residence and beliefs about sibling kinship were all entered into a
multiple regression, MPA was the only variable to independently
predict variance in altruism towards younger siblings (partial
r50.27, P50.001 (0.0013); tolerances, 0.42, 0.54 and 0.50, respec-
tively). Moreover, MPA predicts altruism towards younger siblings
better than either of its component parts (having the same
mother 1sibling coresidence beginning at the sibling’s birth; see
Supplementary Information section 3).
Although the sample size was much smaller (N530), the same
MPA–coresidence pattern emerged for the moral wrongness judg-
ments for incest (Fig. 3). For subjects with one opposite sex younger
sibling, MPA predicted moral opposition at r50.31 (P50.05
(0.06)), about the same effect size as for altruism. When the effects
of coresidence were statistically removed, the effect size for MPA
remained virtually unchanged: r50.26. In contrast, the effect size
for coresidence in predicting moral opposition was low (r50.18,
P50.17 (0.21)), and when the effects of MPA were statistically
removed, it disappeared entirely (r50.01, P50.49 (0.61); tol-
Taken together, these analyses indicate that MPA is indeed a cue
used by olders in detecting younger siblings; when MPA is present,
coresidence duration is no longer used.
Is coresidence a kin cue or an artefact? When MPA is absent, cor-
esidence duration correlates with altruism to the same degree regard-
less of the sibling’s sex, as kin selection theory predicts that a cue to
genetic relatedness should. But individuals are at risk for incest only
from opposite sex siblings. Tellingly, moral opposition to third party
sibling incest tracks duration of coresidence with an opposite sex, but
not a same sex sibling (r520.01, P50.47 (0.59), N530). This
pattern rules out any counter-hypothesis that coresidence duration
is important not because it cues genetic relatedness, but because it is a
spurious correlate of something else about the family (stability, tra-
ditional family structure, religion, and so on)
The effects of coresidence when MPA is absent are also much
targeted: duration of coresidence does not predict generosity outside
of the sibling pair, and it is not positively correlated with moral
judgments about any surveyed behaviours unrelated to incest
(Supplementary Information section 4).
Altruism: effect size (r)Incest moral: effect size (r)
When MPA and coresidence duration cues are both available,
the kin detection system defaults to MPA, the more reliable cue. a,b, The
only individuals for whom these cues could be jointly available are olders
detecting younger siblings; each bar on the graph shows the size of the
correlation between a cue and an outcome measure for this group. For olders
responding to youngers, exposure to the MPA cue predicts both altruism
(a) and moral opposition to sibling incest (b), and with the same effect size
(black bar, first pair, each panel). The MPA cue continues to predict these
disparate measures even after the effects of coresidence duration are
statistically removed (black bar, second pair, each panel). In contrast,
coresidence duration ceases to predict either altruism or moral opposition to
sibling incest once the effects of MPA are removed (grey bar, second pair,
each panel). ***P,0.001, **P,0.01, *P50.05, 1P,0.10.
15 February 2007 ARTICLES
©2007 Publishing Group
Early imprinting? Despite claims for an early imprinting period for
, when MPA is absent, total duration of coresi-
dence predicts altruism and sexual aversion better than age of sibling
(or subject) when coresidence begins (start age; Supplementary
Information section 5). The discovery that MPA is a potent cue for
olders detecting younger sibs might explain past results suggesting an
early imprinting period: start age at sibling’s birth is not an inde-
pendent predictor for olders detecting youngers (Supplementary
Information section 3), but it is one component of the MPA cue.
Do beliefs matter? Coresidence duration predicted the outcome
measures better than subjects’ consciously held beliefs about siblings’
genetic relatedness. Controlling for beliefs, coresidence continued to
predict most outcome measures; in contrast, beliefs failed to predict
most measures once the effects of coresidence were controlled for
statistically (Supplementary Information section 6). Indeed, when
subjects believe their sibling is step or adoptive, coresidence predicts
altruism and sexual aversions, indicating that when beliefs conflict
with the kin detection system, the criteria used by the kin detection
system prevail (Supplementary Information section 6).
Other alternatives? Caution is always warranted in interpreting cor-
relational findings, but it seems safe to say that altruism and sexual
aversion are either regulated by the theoretically predicted cues, MPA
and coresidence duration, or by unidentified cues very highly corre-
lated with them. So far, we have been unable to find any cues that
predict outcomes better than do MPA and coresidence duration.
The tight mesh between theoretical expectations and empirical tests
provides strong support for the hypothesis that humans have a sys-
tem designed by selection to detect genetic relatedness: specifically,
one with (at a minimum) the computational elements outlined in
Fig. 1. For example, the fact that different motivational systems are
regulated in parallel by the same cues to relatedness implicates a
single underlying neurocomputational variable—a kinship index—
used by both. Moreover, if registered information about MPA and
coresidence were fed directly into programs regulating altruism and
sexual aversion, their effects would only be additive. They were not.
Instead, the presence of MPA eliminated effects of coresidence. This
is strong evidence for the existence of an intermediate computational
device, the kinship estimator, equipped with procedures that com-
bine these cues in a non-compensatory
way to compute the kinship
index. These results contribute to a growing body of findings showing
that humans are not immune to the evolutionary forces that have
shaped other species, and that Darwinism has a central role in dis-
covering the neural and psychological architecture of our species.
All subjects completed a survey about family composition and attributes. For
each sibling, subjects indicated that sibling’s age, type of sibling (for example,
biological, step), coresidence duration, age range of coresidence, and certainty of
sharing the same biological mother and father
. From these, the following
predictor variables were constructed: coresidence (duration of time a subject
co-resided with his/her sibling between the subject’s ages of 0 and 18); and
Maternal Perinatal Association (MPA; where a score of 1 means the subject
began coresidence with a sibling at the sibling’s birth and is certain they share
the same biological mother, and a score of 0 means any other scenario).
Instrument 1: sibling-directed altruism. Subjects (N5154 (107 women); ages,
16–21, mean age 6s.d. of 18.4460.82; 287 sibling pairs) indicated the number
of favours they performed for each sibling in the last month (behavioural mea-
sure), and, separately, how willing they would be to donate a kidney to their
sibling (dispositional measure) on a 7-point Likert-like scale (0, not willing at all;
6, extremely willing). Responses from these measures produced the same pattern
of results (Fig. 2) and were summed to produce a dependent variable, altruism
(range, 0 to 16; mean 6s.d. of 7.57 62.83).
Instrument 2: moral wrongness associated with third party sibling incest.
Subjects (N5186 (102 women); ages 18–47, mean 6s.d. of 21.54 64.21)
ranked 19 social transgressions on moral wrongness
. Two acts regarding third
party sibling incest (‘consensual sex between a brother and sister’ and ‘brother–
sister marriage’) were summed to produce a dependent variable, moral opposi-
tion (reverse-coded; range of 7 to 31 (mean 6s.d. of 22.43 65.12)). This vari-
able measures how morally wrong subjects view sibling incest among third
parties (not incest with a particular sibling); therefore, to isolate effects to a
particular sibling (in contrast to analyses in ref. 28), data analysis was restricted
to individuals with only one opposite sex sibling (N574).
Instrument 3: disgust imagining sexual acts with a sibling (Likert). Subjects
(N5455 (264 women); ages 18–54, mean6s.d. of 21.28 63.91; a subset also
completed Instruments 2 and 4) were asked how disgusting they would find
engaging in various sexual and nonsexual behaviours on a 7-point Likert-like
scale (0, not disgusting at all; 6, extremely disgusting). Among these were sexual
acts with particular opposite sex siblings. For each opposite sex sibling, inde-
pendent ratings for passionately kissing, and having sex with ‘your sibling’ were
summed to produce a dependent variable, sexual disgust (Likert).
Initial analyses, for which non-independence was not a concern (see
Supplementary Information section 8), indicated that women were at ceiling
for this measure and showed significantly less variance than men in their res-
ponses (Levine’s F
). The multi response permuta-
tion procedure (MRPP)
indicated that, as predicted, women reported more
disgust at sex with a sibling than did men (women (mean 6s.d.) 11.72 60.98,
N5264; men 11.12 61.96, N5191; standardized test statistic of 212.72,
). For this reason
this variable permitted the exploration of disgust
responses in males, but not females (N5191 males; ages 18–54, mean 6s.d. of
21.09 63.30; 246 sibling pairs).
Sexual disgust (Likert) was transformed into a dichotomous variable: ‘1’ was
assigned if a male responded at ceiling for disgust associated with sex and kissing
a sibling; ‘0’ if otherwise (mean 50.73, s.d. 50.45). For the other three depend-
ent measures, there were no sex differences in the relationships between pre-
dictor and outcome variables so results are reported for men and women
Instrument 4: disgust imagining sexual acts with a sibling (rank). A subset of
participants who completed Instrument 3 also completed Instrument 4
(N5375), which asked participants to assign a unique rank of disgust from 1
(not disgusting at all) to 50 (extremely disgusting) to eight acts, some of which
involved sexual contact with a family member, short of intercourse. Using
the rank of the sexual act involving a sibling, a variable, sexual disgust (rank),
was constructed (women, mean 547.36, s.d. 53.99; men, mean 545.51,
s.d. 59.91). To assess the effects of coresidence on sexual disgust in a way that
reflects coresidence with a particular sibling, data analyses are limited to subjects
with only one opposite sex sibling (N5243 (144 women); ages 18–50,
mean 6s.d. of 21.02 62.95).
Data analyses. Correlations involving dependent measures ‘moral opposition’
and ‘sexual disgust’ controlled for the subject’s sexual orientation. Controlling
for social desirability yielded similar effect sizes. For univariate analyses, we used
directed tests to assess predicted effects
. Pearson correlations for which we had
prior predictions report one-tailed P-values, followed by directed P-values in
parentheses (see Supplementary Information section 9). Non-independence
occurs in Instruments 1 and 3 because some subjects have multiple siblings thus
contributing multiple data-points. For these two studies, separate analyses using
only one sibling pair per subject were carried out and yielded the same effect sizes
(see Supplementary Information section 8).
Received 22 July; accepted 5 December 2006.
1. Hamilton, W. D. The genetical evolution of social behaviour. I, II. J. Theor. Biol. 7,
2. Williams, G. C. & Williams, D. C. Natural selection of individually harmful social
adaptations among sibs with special reference to social insects. Evolution 11,
3. Strassmann, J. E., Zhu, Y. & Queller, D. C. Altruism and social cheating in the social
amoeba Dictyostelium discoideum.Nature 408, 965
4. Crozier, R. H. & Pamilo, P. Evolution of social insect colonies: Sex allocation and kin-
selection (Oxford Univ. Press, Oxford, 1996).
5. Chapuisat, M. & Keller, L. Testing kin selection with sex allocation data ineusocial
hymenoptera. Heredity 82, 473
6. Passera, L., Aron, S., Vargo, E. L. & Keller, L. Queen control of sex ratio in fire ants.
Science 293, 1308
7. Duffy, J. E. Eusociality in a coral-reef shrimp. Nature 381, 512
8. Baglione, V., Canestrari, D., Marcos, J. & Ekman, J. Kin selection in cooperative
alliances of carrion crows. Science 300, 1947
9. Ito, Y. The evolutionary biology of sterile soldiers in aphids. Trends Ecol. Evol. 4,
10. Queller, D. C. Inclusive fitness in a nutshell. Oxford Surveys Evol. Biol. 6, 73
11. Cosmides, L. & Tooby, J. Cytoplasmic inheritance and intragenomic conflict. J.
Theor. Biol. 89, 83
12. Sherman, P. W. Nepotism and the evolution of alarm calls. Science 197, 1246
15 February 2007
©2007 Publishing Group
13. Buchan, J. C., Alberts, S. C., Silk, J. B. & Altmann, J. True paternal care in a multi-
male primate society. Nature 425, 179
14. Chapais, B. & Berman, C. M. (eds) Kinship and Behavior in Primates (Oxford Univ.
Press, New York, 2004).
15. Silk, J. B. Kin selection in primate groups. Int. J. Primatol. 23, 849
16. Fletcher, D. & Michener,C. (eds) Kin Recognition in Animals (Wiley, New York, 1987).
17. Hepper, P. G. Kin Recognition (Cambridge Univ. Press, New York, 1991).
18. Holmes, W. The early history of Hamiltonian-based kin recognition research
theory: past and future. Ann. Zool. Fennici 41, 691
19. Charlesworth, B. & Charlesworth, D. The genetic basis of inbreeding depression.
Genet. Res. 74, 329
20. Crnokrak, P. & Roff, D. A. Inbreeding depression in the wild. Heredity 83, 260
21. Bittles, A. H. & Neel, J. V. The costs of human inbreeding and their implications for
variation at the DNA level. Nature Genet. 8, 117
22. Tooby, J. Pathogens, polymorphism, and the evolution of sex. J. Theor. Biol. 97,
23. Penn, D. J. & Potts, W. K. The evolution of mating preferences and major
histocompatibility coupled genes. Am. Nat. 153, 145
24. Lee, R. B. & Devore, I. Man the Hunter (Aldine, Chicago, 1968).
25. Howell, N. Demography ofthe Dobe!Kung 2ndedn (AldineTransaction, NewYork,
26. Hill, K. & Hurtado, A. Ache Life History: The Ecology and Demography of a Foraging
People (Aldine Transaction, New York, 1996).
27. Bevc, I. & Silverman, I. Early separation and sibling incest: A test of the revised
Westermarck theory. Evol. Hum. Behav. 21, 151
28. Lieberman, D., Tooby, J. & Cosmides, L. Does morality have a biological basis? An
empirical test of the factors governing moral sentiments regarding incest. Proc. R.
Soc. Lond. B 270, 819
29. Fessler, D. M. T. & Navarrete, C. D. Third-party attitudes toward sibling incest:
Evidence for Westermarck’s hypotheses. Evol. Hum. Behav. 25, 277
30. Wedekind, C. & Furi, S. Body odour preferences in men and women: do they aim
for specific MHC combinations or simply heterozygosity? Proc. R. Soc. Lond. B
31. Ober, C. et al. HLA and mate choice in humans. Am. J. Hum. Genet. 61, 497
32. DeBruine, L. M. Trustworthy but not lust-worthy: Context-specific effects of facial
resemblance. Proc. R. Soc. Lond. B 272, 919
33. Westermarck, E. A. The History of Human Marriage 5th edn (Macmillan, London,
34. Wolf, A. P. Sexual Attraction and Childhood Association: A Chinese Brief for Edward
Westermarck (Stanford Univ. Press, Stanford, California, 1995).
35. Shepher, J. Mate selection among second generation kibbutz adolescents and
adults: incest avoidance and negative imprinting. Arch. Sex. Behav. 1, 293
36. Kaplan, H. et al. A theory of human life history evolution: Diet, intelligence, and
longevity. Evol. Anthropol. 9, 156
37. Hewlett, B. & Lamb, M. Hunter
Gatherer Childhoods (Aldine Transaction,
Somerset, New Jersey, 2005).
38. Walters, J. R. in Kin Recognition in Animals (edsFletcher, D. J. C.& Michener, C. D.)
393 (Wiley & Sons, New York, 1987).
39. Bernstein, I. in Kin Recognition (ed. Hepper, P. G.) 6
29 (Cambridge Univ. Press,
40. Gigerenzer, G., Todd, P., ABC Research Group. Simple Heuristics That Make Us
Smart (Oxford Univ. Press, New York, 1999).
41. Mielke, P. W. & Berry, K. J. Permutation Methods: A Distance Function Approach
(Springer, New York, 2001).
42. Cade, B. S. & Richards, J. D. User Manual for BLOSSOM Statistical Software
(Midcontinent Ecological Science Center, US Geological Survey, Fort Collins,
43. Rice, W. R. & Gaines, S. D. Heads I win, tails you lose: Testing directional
alternative hypotheses in ecological and evolutionary research. Trends Ecol. Evol.
Supplementary Information is linked to the online version of the paper at
Acknowledgements The authors thank P. Boyer, D. Fessler, S. Gangestad,
P. Pocker, H. Waldow, G. Williams, D. Williams, UCSB Academic Senate and the
providers of the NSF Presidential Young Investigator Award (J.T.), and NIH
Director’s Pioneer Award (L.C.).
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Correspondence and requests for materials should be addressed to D.L.
15 February 2007 ARTICLES
©2007 Publishing Group