Resolving the paradox of common, harmful, heritable mental disorders:

Resolving the paradox of common, harmful, heritable mental disorders:
Printed in the United States of America
Resolving the paradox of common,
harmful, heritable mental disorders:
Which evolutionary genetic models
work best?
Matthew C. Keller
Virginia Institute for Psychiatric and Behavioral Genetics, Virginia
Commonwealth University, Richmond, VA 23219.
[email protected]
Geoffrey Miller
Department of Psychology, University of New Mexico, Albuquerque,
NM 87131-1161.
[email protected]
Abstract: Given that natural selection is so powerful at optimizing complex adaptations, why does it seem unable to eliminate genes
(susceptibility alleles) that predispose to common, harmful, heritable mental disorders, such as schizophrenia or bipolar disorder? We
assess three leading explanations for this apparent paradox from evolutionary genetic theory: (1) ancestral neutrality (susceptibility
alleles were not harmful among ancestors), (2) balancing selection (susceptibility alleles sometimes increased fitness), and
(3) polygenic mutation-selection balance (mental disorders reflect the inevitable mutational load on the thousands of genes
underlying human behavior). The first two explanations are commonly assumed in psychiatric genetics and Darwinian psychiatry,
while mutation-selection has often been discounted. All three models can explain persistent genetic variance in some traits under
some conditions, but the first two have serious problems in explaining human mental disorders. Ancestral neutrality fails to explain
low mental disorder frequencies and requires implausibly small selection coefficients against mental disorders given the data on the
reproductive costs and impairment of mental disorders. Balancing selection (including spatio-temporal variation in selection,
heterozygote advantage, antagonistic pleiotropy, and frequency-dependent selection) tends to favor environmentally contingent
adaptations (which would show no heritability) or high-frequency alleles (which psychiatric genetics would have already found).
Only polygenic mutation-selection balance seems consistent with the data on mental disorder prevalence rates, fitness costs, the
likely rarity of susceptibility alleles, and the increased risks of mental disorders with brain trauma, inbreeding, and paternal age.
This evolutionary genetic framework for mental disorders has wide-ranging implications for psychology, psychiatry, behavior
genetics, molecular genetics, and evolutionary approaches to studying human behavior.
Keywords: adaptation; behavior genetics; Darwinian psychiatry; evolution; evolutionary genetics; evolutionary psychology; mental
disorders; mutation-selection balance; psychiatric genetics; quantitative trait loci (QTL)
1. Introduction
Mental disorders such as schizophrenia, depression,
phobias, obsessive-compulsive disorder, and mental
retardation are surprisingly prevalent and disabling. In
industrialized countries such as the United States, an estimated 4% of people have a severe mental disorder
(National Institute of Mental Health 1998), and almost
half of people will meet the criteria for some type of less
severe mental disorder at some point in their lives
(Kessler et al. 2005). The annual economic costs in treatment and lost productivity are in the hundreds of billions
of dollars (Rice et al. 1992). The less quantifiable personal
costs of mental disorders to sufferers, families, and friends
are even more distressing. For example, schizophrenia
affects about 1% of people worldwide (Jablensky et al.
1992), typically beginning in early adulthood and often
following a chronic lifelong course. People with
# 2006 Cambridge University Press
0140-525X/06 $12.50
schizophrenia often imagine hostile, confusing voices;
they have trouble thinking clearly, feeling normal
emotions, or communicating effectively; and they tend to
lose jobs, friendships, and sexual partners. In response,
many people with schizophrenia kill themselves, and a
much larger proportion dies childless.
This is an evolutionary puzzle, because differences in
the risk of developing schizophrenia and other common,
debilitating mental disorders are due, in large part, to
differences in people’s genes. Given that natural selection
has built the most exquisitely complex machinery known to
humankind – millions of species of organic life-forms –
why do so many people suffer from such debilitating and
heritable mental disorders? If these mental disorders are
as disabling as they appear, natural selection should have
eliminated the genetic variants (susceptibility alleles)
that predispose to them long ago. Does the prevalence
of heritable mental disorders therefore imply that mental
Keller & Miller: Resolving the paradox of heritable mental disorders
disorder susceptibility alleles were selectively neutral or
perhaps even advantageous in the ancestral past, or has
natural selection been unable to remove susceptibility
alleles for some hidden reason?
1.1. The goal of this article and who should read it
This article tries to develop an understanding of the evolutionary persistence of susceptibility alleles underlying
common, heritable, harmful mental disorders. We
compare and contrast the three broadest classes of evolutionary genetic models that explain persistent genetic
variation: ancestral neutrality, balancing selection, and
polygenic mutation-selection balance. Such models have
been tested mostly by evolutionary geneticists on traits
such as bristle numbers in fruit flies, survival in nematode
worms, and growth rates in baker’s yeast. Yet these models
make strong, discriminating predictions about the genetics, phenotypic patterns, and fitness payoffs of any trait
in any species, and so should be equally relevant to
explaining mental disorder susceptibility alleles.
However, these three main models of persistent genetic
variation have never before been directly compared with
regard to their theoretical and empirical adequacy for
explaining human mental disorders. That is our first
main goal.
Our second main goal is to promote more consilience
among evolutionary genetics, human behavioral/psychiatric genetics, and Darwinian psychiatry/evolutionary
psychology. Trying to integrate these disparate fields is
hard, not just because each field has different goals,
terms, assumptions, methods, and journals, but also
because each field has various outdated misunderstandings of one another. For example, we will argue that
Darwinian psychiatry often relies too heavily on balancing
selection, whereas psychiatric genetics often assumes
fitness neutrality or ignores evolutionary forces altogether.
Although balancing selection and neutral evolution were
historically seen as primary causes of genetic variation,
MATTHEW C. KELLER is a postdoctoral fellow at the
Virginia Institute for Psychiatric and Behavioral
Genetics. He received a B.A. from the University of
Texas, Austin in 1995 and a Ph.D. from the University
of Michigan, Ann Arbor in 2004. He has done postdoctoral work in genetic epidemiology at the Queensland
Institute of Medical Research in Brisbane, Australia,
and at the Center for Society and Genetics at UCLA.
His primary interests are in behavioral/psychiatric
genetics, evolutionary psychology, evolutionary theory,
personality, emotion, and statistical methods.
GEOFFREY MILLER is an evolutionary psychologist at
the University of New Mexico. He received a B.A.
from Columbia University in 1987, and a Ph.D. from
Stanford University in 1993, then worked in Europe
until 2001 (at University of Sussex, University College
London, London School of Economics, and the Max
Planck Institute for Psychological Research in
Munich). His book The Mating Mind (2000) has been
published in 11 languages. His research concerns
human mate choice, fitness indicators, evolutionary
behavior genetics, intelligence, personality, psychopathology, and consumer behavior.
they have proven less important than expected in explaining persistent genetic variation in traits related to fitness.
Conversely, the third model – polygenic mutationselection balance – has enjoyed a theoretical and empirical
renaissance in evolutionary genetics, but remains
obscure and misunderstood in psychiatric genetics and
Darwinian psychiatry.
Cross-fertilization between these fields promises not
only to shed light on deep quandaries regarding the
origins of mental disorders; it also may help resolve
some ongoing frustrations within each field by guiding
research and theory more effectively. Evolutionarily
oriented mental health researchers, such as Darwinian
psychiatrists and evolutionary psychologists, often go to
torturous lengths to find hidden adaptive benefits that
could explain the evolutionary persistence of profoundly
harmful mental disorders such as schizophrenia or anorexia, but these accounts are often frustratingly implausible
or hard to test. New ideas from evolutionary genetics and
data from psychiatric genetics can help this audience
better understand which evolutionary genetic models are
theoretically credible and empirically relevant to mental
Many psychiatric and behavioral geneticists try to find
the specific susceptibility alleles that underlie common,
harmful, heritable mental disorders. They are often frustrated that even the most promising loci explain little
overall population risk and rarely replicate across studies
or populations. Traditional methods for gene hunting
implicitly assume that mental disorder susceptibility
alleles will be at relatively high frequencies and common
across populations. Such a convenient scenario, we will
argue, could arise from ancestral neutrality or balancing
selection, but is much less likely to arise from a
mutation-selection balance. Evolutionary genetics could
help guide more fruitful gene hunting based on more
realistic assumptions.
Evolutionary geneticists try to understand the origins
and implications of natural genetic variation across traits
and species. The beautiful empirical and theoretical
work in evolutionary genetics is under-funded and too
often thought irrelevant to human welfare. Greater familiarity with evolutionary genetics might help funding
agencies appreciate the potential relevance of this work
to understanding some of the leading causes of human
suffering, and may introduce evolutionary geneticists to
rich genetic data sets on complex human traits such as
mental disorders that can be used to test evolutionary
1.2. What this article owes to Darwinian psychiatry
In developing our ideas, we build upon Darwinian psychiatry as it has developed over the last 20 years
(McGuire & Troisi 1998; Nesse & Williams 1994;
Stevens & Price 2000a). Our starting point is the Darwinian psychiatric view that dysfunction is difficult to infer
without an understanding of function (Troisi & McGuire
2002; Wakefield 1992). Mental disorders, by this viewpoint, reflect a failure of one or more psychological adaptations to perform their proper, naturally selected,
prehistoric functions (Troisi & McGuire 2002; Wakefield
1992). The heart is an adaptation designed to pump
blood, for example, and its failure causes blood-circulation
Keller & Miller: Resolving the paradox of heritable mental disorders
problems that are functionally distinguishable from a
pancreas’s failure to regulate blood sugar or a lung’s
failure to oxygenate blood. Likewise, there is a clear
mental health problem when a brain is unable to feel
social emotions or make sense of reality. This perspective
has some important corollaries.
First, a better understanding of normal psychological
adaptations should help delineate harmful dysfunctions
in those adaptations. Research on adaptive function (e.g.,
evolutionary psychology; Barkow et al. 1992; Buss 1995)
and research on maladaptive dysfunction (e.g., Darwinian
psychiatry) are mutually illuminating. This is equally true
when mental disorder symptoms have only indirect
relationships to psychological adaptations. For example,
reading disorders cannot result from a dysfunction in a
“reading adaptation,” because the visual and linguistic
adaptations that enable reading evolved long before the
invention of writing a few thousand years ago (Wakefield
1999a). Likewise, auditory hallucinations in schizophrenia
probably do not result directly from dysfunction in a
“hallucination-suppression adaptation,” but indirectly, as
side-effects of dysfunctions in more plausible mechanisms
that, for example, coordinate and store short-term
information, or that filter irrelevant stimuli (Cannon &
Keller 2005).
Second, many mental disorders are probably extreme
points along a continuum of symptom severity that
ranges from patently unaffected to extreme forms of the
disorder. This makes distinctions between “normal” and
“abnormal” somewhat arbitrary (Farmer et al. 2002),
because psychological adaptations often show continuous
degradation of performance. In this dimensional view of
mental disorders, schizophrenia is an extreme form of
schizotypal and schizoaffective personality disorders,
mental retardation is an extreme form of low intelligence,
chronic depression is an extreme form of normal depressive reactions, and so forth. Even mental disorders that look
like discrete categories at the phenotypic macro-level
(mainly eating, dissociative, post-traumatic stress, melancholic depressive, and antisocial disorders; Haslam 2003)
may be influenced by the cumulative effect of many
minor dysfunctions at the micro-level of genes and brain
development (Gottesman & Shields 1967).
Third, some apparently pathological behaviors may not
really be disorders at all from an evolutionary perspective
because they do not reflect genuine maladaptive dysfunctions. In particular, some clinically defined mental disorders such as certain phobias or depressions may be
reactive defenses analogous to fever, nausea, and bodily
pain, which protect against infections, toxins, and tissue
damage, respectively (Gilbert 1998; McGuire & Troisi
1998; Nesse & Williams 1994). Aversive defenses are
cues that something in the environment is wrong, not
pathologies themselves.
Consider, for example, depression in light of the reactive defense model. In response to major failures or
losses, normally expressed depressive symptoms (e.g.,
pessimism and fatigue) may adaptively withdraw effort
from unpropitious situations when the marginal fitness
returns are likely to be low, and emotional pain may motivate avoidance of such situations in the future (Keller &
Nesse 2005; 2006; Nesse 2000). These normal reactions
are illustrated by the regression line in Figure 1; more
severe situations provoke more protracted and severe
Figure 1. The reactive defense model as applied to depressive
reactions (see T.A. text).
reactions. The Gaussian distributions in Figure 1 illustrate
interpersonal differences, including genetic differences,
which influence symptom severity, given a certain level
of situation severity. Since severe situations (e.g., major
failures, death of kin) can cause nearly anyone to experience depressive symptoms (Monroe & Simons 1991),
some cases of severe and prolonged depressive symptoms
(i.e., clinical depression; see above the dashed line in
Fig. 1) may simply be normal and adaptive responses to
very adverse situations. At the same time, depressive
symptoms that are abnormally severe, given the situation
(the positive extremes of the Gaussian distributions),
may signify malfunctions in the mechanisms responsible
for depressive symptoms. Thus, clinical cut-offs based
solely on symptom severity and duration, and which do
not consider the fitness-relevant precipitating situation,
may fail to distinguish truly pathological from nonpathological depressive symptoms (Wakefield 1999a).
These insights are generally appreciated in Darwinian
psychiatry and eventually should help build a comprehensive theoretical framework for psychiatry. However, there
remains a gaping hole in Darwinian psychiatry’s account of
mental disorders: there are no good explanations of why
human brains seem to malfunction so often, and why
these malfunctions are both heritable and disastrous to
survival and reproduction. That is, there is still no good
answer for why such susceptibility alleles have persisted
despite thousands of generations of natural selection for
adaptive human behavior.
1.3. What phenomena this article tries to explain
This article tries to develop an understanding of the evolutionary persistence of susceptibility alleles: regions of
DNA – broadly defined to include both coding as well
as non-coding, regulatory regions (see sect. 6.3) – that
differ between individuals in the population and that
increase the risk of common mental disorders. In other
words, this article is concerned with explaining the
genetic rather than the environmental variation associated
with mental disorders (with complications such as geneenvironment interactions considered later [sect. 4.4]).
The reactive defense model offers insight into the environmental triggers for certain disorders – the normal
Keller & Miller: Resolving the paradox of heritable mental disorders
reactions to environmental stressors and the suites of
species-typical, fixed alleles that code for these reactions.
However, the reactive defense model is not helpful in
explaining genetic variation, because adaptive defenses
should be activated by environmental triggers, not heritable risk alleles that differ between individuals.
Although we have continually referred to mental disorders such as schizophrenia, depression, or mental retardation as “common,” these disorders are uncommon in an
absolute sense, generally having lifetime prevalence rates
of less than 2%. Rather, they are common relative to the
thousands of other heritable states that are known to be
harmful to fitness, such as achondroplastic dwarfism or
Apert’s syndrome.
Most rare, harmful, single-gene disorders (Mendelian
disorders) have frequencies consistent with mutationselection balance – a balance between genetic copying
errors that turn normal alleles into harmful mutations,
and selection eliminating these mutations (Falconer &
Mackay 1996). Mutations arise in parental germ-line
cells and are passed on to offspring (and all their cells,
including their own germ-line cells) at some low rate (m)
per gene, per individual, per generation. Those that
affect the phenotype are almost always harmful for the
same reason that random changes to a computer’s circuitry
are almost always harmful: entropy erodes functional complexity (Ridley 2000). Selection removes these mutations
at a rate proportional to the fitness cost of the mutation,
represented by the selection coefficient (s) against the
mutation. If s is reproductively lethal (s ¼ 1), the newly
arisen mutation exists in only one body before being eliminated from the population, but if s is fairly small, the
mutation may pass through and affect many bodies
through many generations before being removed by selection. The result of this balance between mutation rate m
and selection coefficient s is usually a low equilibrium
frequency (p) of mutant alleles that have not yet been
removed from the population by selection. Specifically,
mutations are expected to have
pffiffiffiffiffiffiffiffipopulation frequencies
of p ¼ m/s if dominant, p ¼ m=s if recessive, and somewhere in between otherwise (additive alleles are exactly
midway between). As mutation rate m decreases or
selection coefficient s increases, the mutation’s frequency
p should drop. This process accurately describes, in most
cases, why Mendelian disorders are so rare.
The cumulative frequency of all Mendelian disorders –
around 2% of all births (Sankaranarayanan 2001) – is
high only because so many genes are subject to mutation
(around 25,000). Heritable harmful disorders that are individually this rare (,1/5,000) pose no evolutionary
paradox; no one wrings their hands about trying to find
hidden adaptive benefits for such disorders because their
frequencies are consistent with a simple balance between
mutation and selection. Thus, a more accurate way to classify mental disorders as “common” or “rare” is to assess
whether they are much more common than would be
expected from a single-gene mutation-selection balance.
Table 1 compares the frequencies of several mental
disorders with the frequencies of several Mendelian disorders, all of which are consistent with mutation-selection
expectations (except for sickle-cell anemia, discussed in
sect. 5.4). Stunningly, common mental disorders tend to
be hundreds and even thousands of times more prevalent
than expected from a single-gene mutation-selection
model. This discrepancy has led many researchers (e.g.,
D. R. Wilson 1998) to dismiss mutation-selection
Table 1. Comparisons of frequencies between a small subset of Mendelian disorders
and common mental disorders
Mendelian disorders
Dyskeratosis congenita
Granulomatous disease, type I
Apert’s syndrome
Juvenile onset Parkinson’s
Achondroplastic dwarfism
Sickle-cell anemia
Common mental disorders
Tourette’s syndrome
Anorexia nervosa
Bipolar disorder
Mild mental retardationb
Obsessive-compulsive disorder
Panic disorders
Genetic basis
Lifetime prevalence per
100,000 in U.S.A.
Recessive mutations at 3q25
Recessive mutations at 7q11.23
Dominant mutations at 10q26
Recessive mutations at 1p & 6q26
Dominant mutations at 4q
Recessive mutation at 11p15.5
h 2 ffi .90
h 2 ffi .90
h 2 ffi .65
h 2 ffi .60
h 2 ffi .80
h 2 . .65
h 2 ffi .45
h 2 ffi .30
h 2 ffi .45
20 – 50
1,700 – 3,500
5,000 – 17,000
Note: Data obtained from Online Mendelian Inheritance of Man (n.d.) for Mendelian disorders and from the
National Institute of Mental Health (1998) for common mental disorders unless otherwise noted. When single
or best estimates of heritability or prevalence were unavailable, we used the average of the reported estimates.
Among African Americans.
Heritability and prevalence data derived from Vogel and Motulsky (1997).
Keller & Miller: Resolving the paradox of heritable mental disorders
balance as a viable explanation for certain mental disorders, and to doubt that mental disorder susceptibility
alleles were ancestrally maladaptive. However, such a conclusion is unwarranted. While single-gene mutation-selection models can clearly be eliminated as explanations for
the mental disorders listed in Table 1, multiple-gene (polygenic) models (e.g., Shaner et al. 2004) cannot.
This article focuses on the susceptibility alleles of
mental disorders that are much more common than
would be expected from a single-gene mutation-selection
balance; roughly, this corresponds to mental disorders
with lifetime prevalence rates above 50 per 100,000 in
reproductively aged adults. The best-studied of such disorders are listed in Table 1, but we do not attempt to
provide an exhaustive list of precisely what mental disorders this entails, in part because we suspect that the
sundry categories of modern mental disorders are not
very meaningful biologically (see sects. 6 and 8), but also
because our focus is on understanding the persistence of
susceptibility alleles in general rather than on understanding mental disorders individually. Nevertheless, the types
of common mental disorders that pose the largest
paradox are those that are the most harmful (anorexia,
bipolar disorder, schizophrenia, mental retardation, and
obsessive-compulsive disorder). When we refer to
mental disorders, these are the types of disorders we
have in mind. If we can explain the susceptibility alleles
for disorders that are this debilitating, then the same explanations should provide insight into susceptibility alleles for
somewhat less debilitating disorders (panic disorders and
depression). The following section examines the central
paradox of susceptibility alleles in more detail.
2. The paradox of common, harmful, heritable
mental disorders
The complexity, optimality, and diversity of life on Earth
reveal the awesome power of natural selection.
Common, harmful, and heritable mental disorders (as
well as other disorders that are not the focus of the
current article) seem to be glaring exceptions. They pose
an evolutionary paradox because natural selection is
expected to make harmful, heritable traits very uncommon
very quickly. Over evolutionary time, selection favors
higher-fitness alleles; alleles at most genetic loci have
gone to fixation (virtually 100% prevalence) because they
promoted survival and reproduction under ancestral
conditions better than other alleles did on average. Such
alleles comprise the species-typical human genome; its
normal neurodevelopmental product is human nature.
Lower-fitness alleles, on the other hand, even those with
very minor negative effects, tend to go extinct fairly
quickly. Alleles that reach fixation or extinction cause no
genetic variation, and so cannot contribute to heritable
variation in traits, such as mental disorders. This expectation that selection should minimize genetic variation
in fitness-related traits was canonized in evolutionary
theory as a major implication of Fisher’s fundamental
theorem of natural selection (Fisher 1930/1999). For
decades, biologists expected that the stronger the selection
on a trait, the less heritable variation the trait should show,
and early empirical data seemed supportive.
Based on such reasoning, evolutionary psychologists
have usually argued that genetic variation in human
psychological traits is likely to be either adaptively
neutral (e.g., Tooby & Cosmides 1990) or adaptively maintained by balancing selection (e.g., Mealey 1995). Both
explanations require that the alternative alleles underlying
a trait’s heritable variation have net fitness effects that are
exactly equal to each other, when averaged across evolutionary time and ancestral environments. These explanations seem less relevant to mental disorders, which
appear to be the very embodiment of maladaptive traits.
Nevertheless, the expectation that selection knows best,
and that genetic variation in any common trait cannot be
maladaptive, led to something of a cottage industry
among Darwinian psychiatrists trying to explain the evolutionary persistence of alleles that increase the risk of
such mental disorders as schizophrenia (Horrobin 2002;
Huxley et al. 1964; Jarvik & Deckard 1977; Polimeni &
Reiss 2002; Stevens & Price 2000a), bipolar disorder
(Sherman 2001; D. R. Wilson 1998), depression (D.
R. Wilson 2001), and anorexia (Guisinger 2003). In
response, clinicians more familiar with psychiatric
hospitals, prisons, and detox centers were understandably
skeptical that such apparently Panglossian evolutionary
ideas could explain real mental illness (e.g., Brüne 2004;
McCrone 2003).
Can an evolutionary account of mental disorder susceptibility alleles be reconciled with the clinical view of
mental disorders as genuine dysfunctions? Because they
reveal interesting misunderstandings of the problem, we
begin by considering the most commonly invoked nonviable possible evolutionary explanations of mental disorder
susceptibility alleles. We next consider the (chiefly theoretical) merits of three explanations – ancestral neutrality,
balancing selection, and polygenic mutation-selection
balance – that are better grounded in modern evolutionary genetics. We then discuss six pieces of empirical
evidence, concerning the relationships between mental
disorders and fertility, brain trauma, paternal age,
inbreeding, comorbidity, and frequencies and effect sizes
of mental disorder susceptibility alleles that help
distinguish between these explanations. We conclude
with implications for future research.
3. Non-resolutions to the paradox of common,
harmful, heritable mental disorders
3.1. Mental disorders are not really heritable
After decades of consistent behavioral genetic research,
the hypothesis that genes play no role in mental disorders
(e.g., Ross & Pam 1995) is simply no longer tenable. Using
different methodologies, behavioral geneticists have
consistently found that mental disorder heritability
estimates range from about .2 to about .8, meaning that
20% to 80% of the differences between individuals in
mental disorder liability are accounted for by differences
in alleles between people. Without acknowledging
genetic influences on mental disorders, only the most
convoluted, post hoc arguments could explain why (a)
adopted children are consistently more similar to their
biological than to their adoptive parents, (b) siblings and
twins reared apart are about as similar as siblings and
twins reared together, (c) similarity in extended families
Keller & Miller: Resolving the paradox of heritable mental disorders
decreases monotonically as a function of genetic similarity,
and (d) identical twins are consistently more similar than
fraternal twins (Bouchard et al. 1990; Plomin et al. 2001).
Three issues regarding mental disorder heritability estimates do merit clarification, however. First, heritability
describes how much genetic or environmental factors
play a role in causing differences in a trait; it tells us
nothing about the causes of similarities in a trait. Both
environmental and genetic factors are 100% necessary
for the species-typical expression of every trait, including
every mental disorder. While true, this fact does not
provide an answer to why alleles that create differences in
mental disorder risk persist. Second, finding positive heritability for a mental disorder does not vindicate the mental
disorder as a diagnostic category. To a first approximation,
every reliably measured behavioral trait shows positive
heritability – even constructs such as television viewing
(Plomin et al. 1990) and political attitudes (Eaves et al.
1999). Any arbitrary “disorder” composed of unrelated
but heritable symptoms will show credible heritability.
Last, heritability is a statistical construct that averages
over a lot of complexity. The causal pathways between
genes and the heritable behaviors they influence must
be mediated by many factors, both genetic and environmental in nature. If these factors differ across populations,
cohorts, or environmental conditions, then heritability
estimates – and even the specific genes responsible for
the heritability – might also differ across populations,
cohorts, or environmental conditions. For example, if
body size is associated with successful aggression in one
particular society, then genes that normally influence
size will also influence aggression in that society (this
concept is sometimes called reactive heritability; Tooby &
Cosmides 1990). Thus, in some cases, contemporary
heritability may not accurately reflect ancestral heritability
in magnitude or in composition – a point we consider in
more depth later (sects. 4.2 and 4.4) when discussing
gene-by-environment interactions.
3.2. Mental disorders are not common enough
to hurt the species
One might argue that the cumulative frequency of severe
mental disorders, around 4%, is not high enough to imperil
the survival of the human species. Alternately, one might
argue that the genetic variation underlying mental disorders persists because it is the essential raw material for
future evolutionary progress (Embry 2002). These points
ignore the central lesson of evolutionary genetics: selection acts on competing alleles within a species, without
regard to long-term species viability or evolvability
(Williams 1966). Natural selection is a purely mechanistic
and iterative process whereby alleles from one generation
have a non-random probability of being represented in
subsequent generations. Natural selection does not –
indeed cannot – hedge bets by stockpiling genetic variation in the hope that currently maladaptive alleles
might become adaptive in the future.
3.3. Mental disorders are not really harmful
to individual fitness
It is sometimes argued that mental disorders were not
fitness reducing in ancestral environments because
humans reproduced earlier than they do today (e.g.,
Hardcastle 2004; Weisfeld 2004). However, every mental
disorder in Table 1 strikes well before ancestral humans
would have finished reproducing. A harmful mental disorder that struck even as late as the forties would have
led to a small but evolutionarily significant decrement in
number of future offspring (e.g., see the fertility function
of hunter-gatherers in Daly & Wilson [1983]), even apart
from its negative effect on inclusive fitness through
reduced ability to aid relatives (Kaplan et al. 2000).
Thus, if mental disorders were debilitating in ancestral
conditions, their developmental timing would have
harmed fitness given any reasonable model of ancestral
life-history profiles.
Another version of the not-really-harmful argument
concerns the fitness effects of susceptibility alleles rather
than mental disorders per se: mental disorders may be
harmful to fitness, but their genetic architecture may be
so complex that natural selection has been unable to eliminate the alleles that predispose to them. Used in this
sense, “genetic complexity” basically means nonadditive
genetic variation: variation in fitness effects that depend
on particular combinations of alleles, and that selection
therefore affects at a much slower rate (Merilä &
Sheldon 1999). Such nonadditive effects include
dominance (interactions between two alleles at the same
locus) and epistasis (interactions between alleles at
different loci). However, for the same reasons that main
effects almost always exist in addition to interaction
effects in statistical analyses, dominant and epistatic
alleles almost always have some average, or additive, phenotypic effects (contributing to additive genetic variation)
that are more visible to selection (Falconer & Mackay
1996; Mather 1974). Available empirical evidence on
mental disorders is consistent with this expectation.
Although the vast majority of behavioral genetic studies
have used a design (the classical twin design) that cannot
simultaneously estimate additive, nonadditive, and
shared-environment effects (Eaves et al. 1978; Keller &
Coventry 2005), behavioral genetic studies using designs
better able to distinguish these (such as the extended
twin design; reviewed in Coventry & Keller 2005) have
found at least some additive genetic variation for those
mental disorders investigated to date: depressive symptoms, panic disorders, and neuroticism (a correlate of
many mental disorders). Thus, the harm that mental
disorders do is almost certainly visible to natural selection
to some degree.
It could also be argued that mental disorders simply
have not affected survival and reproduction, and so are
not under selection. At least in modern environments,
however, many mental disorders are associated with markedly lower fertility (summarized in Table 2). These mental
disorders seem to undermine fertility not so much through
reducing survival, but through reducing attractiveness or
ability in the mating arena. Of the studies that examined
this issue, reductions in fertility were principally the
result of lower marriage rates rather than fewer offspring
once married. At this level of socio-sexual competition to
attract and retain mates, there may not be so much difference between the fitness effects of mental disorders in prehistoric and contemporary societies (Miller 2000a).
However, modern fertility has an unknown relationship
to ancestral fertility (Symons 1989), which is more relevant
Table 2. Available fertility estimates (1960 – 2005) of common mental disorders
Birth cohorts, location
58% C
45% C
70% C
40% F; 57% C
29% F; 83% C
23% F; 51% C
101% C
29% F; 62% C
27% F; 45% C
46% C
23% F; 12% C
1890– 1919, U.S.
1890s– 1950s, Germany
1890s– 1940s, U.K.
1911– 1940, U.S.A.
1914– 1968, Spain
1920s– 1970s, Australia
1921– 1976, Canada
1932– 1951, U.S.A.
1930s– 1970s, Japan
1930s– 1970s, Ireland
1950s, Finland
1953– 1982, U.K.
20th century, Denmark
20th century, Palau
4,041 inpatients & outpatients
306 inpatients
1,086 inpatients & outpatients
4,023 inpatients & outpatients
142 inpatients & outpatients
282 primary-care patients
36 primary-care patients
223 outpatients
553 outpatients
285 from population register
11,231 from population register
4,556 primary-care patients
27 from adoption database
70 unknown
Erlenmeyer-Kimling et al. 1969
Vogel 1979
Slater et al. 1971
Erlenmeyer-Kimling et al. 1969
Fananás & Bertranpetit 1995
McGrath et al. 1999
Bassett et al. 1996
Burr et al. 1979
Nanko & Moridaira 1993
Kendler et al. 1993
Haukka et al. 2003
Howard et al. 2002
Rimmer & Jacobsen 1976
Sullivan & Allen 2004
1890s– 1950s, Germany
1890s– 1950s, U.S.A.
1890s– 1940s, U.K.
1920s– 1970s, Australia
1953– 1982, U.K.
165 inpatients
134 inpatients
2,692 inpatients & outpatients
60 primary-care patients
1,705 primary-care patients
Vogel 1979
Baron et al. 1982
Slater et al. 1971
McGrath et al. 1999
Howard et al. 2002
40% F; 72% C
1870s– 1930s, Minnesota
1870s, Michigan
1870s– 1930s, Minnesota
1890s– 1950s, Germany
1,450 descendants of inpatients
78 from school register
1,300 descendants of inpatients
275 inpatients
Reed 1971
Bajema 1963
Waller 1971
Vogel 1979
47% C
64% C
1890s– 1940s, U.K.
1890s– 1940s, U.K.
1890s– 1950s, Germany
235 inpatients & outpatients
5,596 inpatients & outpatients
316 inpatients
Slater et al. 1971
Slater et al. 1971
Vogel 1979
F; 62% C
F; 89% C
Note: Data include all available studies in 1960 – 2005 in which overall fertility rates were reported or derivable and in which a suitable comparison group was reported.
Number of offspring as a proportion of number of offspring among general population matched on age, gender, and other pertinent demographic variables.
Schizophrenia, schizoaffective disorder, schizophreniform, delusional disorder, and paranoid psychosis.
Major depression and bipolar disorder.
Bipolar disorder, bipolar disorder with psychosis, mania, mania with psychosis, and depression with psychosis.
IQ , 70.
IQ , 85.
Mental retardation and psychoses caused by trauma.
Obsessive-compulsive disorder.
Usage not described.
Panic disorder, obsessive-compulsive disorder, drug and alcohol dependence, sexual deviance, and personality disorders.
Keller & Miller: Resolving the paradox of heritable mental disorders
Psychotic disorders
Mood disorders
Affective disorderc
Bipolar disorder
Bipolar disorder
Affective disorderd
Affective disorderc
Developmental disorders
Mental retardatione
Low intelligencef
Mental detardatione
Organic disordersg
Other disorders
Keller & Miller: Resolving the paradox of heritable mental disorders
to understanding the evolutionary persistence of susceptibility alleles. Additional and perhaps more persuasive
evidence that mental disorders were associated with
decreased ancestral fitness is simply based upon the ubiquitous evidence of their deviance and disability in
modern societies, irrespective of their effects on fertility
(Troisi & McGuire 2002; Wakefield 1992). Any psychiatric
book or journal reveals many such examples, which do not
need to be enumerated here. If mental disorders existed in
ancestral environments in much the same form as they do
now, it is reasonable to assume that, at some level of severity, they would have resulted in lower ancestral fitness.
Nevertheless, mental disorders may not have existed in
ancestral environments as they do now. This final version
of the not-really-harmful view merits more careful consideration – the idea that, although mental disorders or their
susceptibility alleles are harmful under modern conditions, they may not have been harmful under ancestral
conditions, when humans lived in small-scale, huntergatherer societies. We assess this hypothesis next.
4. Can ancestral neutrality explain common,
harmful, heritable mental disorders?
It seems unlikely that mental disorder susceptibility alleles
had no effect on ancestral fitness, given that mental disorders are associated with lower fitness (Table 2) and
severe impairment in modern environments. Nevertheless, it is possible that mental disorders were associated
with more benign symptoms or less ostracism ancestrally
so that they were effectively neutral traits. For example,
a common speculation is that perhaps prehistoric individuals with schizophrenia were valued shamans, with a
special social role as religious visionaries, so perhaps
they were not socially and sexually ostracized as in contemporary societies (Polimeni & Reiss 2002; Preti & Miotto
1997). Alternatively, perhaps alleles that increase the risk
of mental disorders today had no such effect in ancestral
environments. From an extended-phenotype perspective,
both cases are examples of gene-by-environment (G – E)
interactions, which occur when the effects of alleles
differ depending upon the physical or social environment.
Is it possible that the fitness effects of mental disorder
susceptibility alleles were equal to the fitness effects of
non-susceptibility alleles in ancestral environments,
enabling them to persist?
4.1. Neutral evolution maintains genetic variation only
when combined with recurrent neutral mutation
To assess whether ancestral neutrality is a viable explanation for the persistence of mental disorder susceptibility
alleles, we must first understand the conditions under
which neutrality maintains genetic variation. The frequencies of neutral alleles are governed by genetic drift – random
sampling error over evolutionary time. Over the long
term, drift leads to genetic uniformity because neutral
alleles either fixate or are lost through sampling error.
Drift almost never maintains neutral alleles at intermediate frequencies where they could explain heritable variation in mental disorder susceptibility. Drift is stronger
in smaller populations, such as ancestral hominid populations, which are more susceptible to sampling error.
Without some additional force that either replenishes
lost alleles (see the next paragraph) or that counteracts
the process of random drift (see the next section), one
neutral allele eventually fixates and the alternative alleles
go extinct.
Depending on the way that new mutations affect mental
disorder risk, recurrent neutral mutations might counteract the loss of genetic variation caused by drift. Mutations
can occur anywhere along a locus, the coding region of
which is typically about 2,000 base pairs long; like lightning, mutations are very unlikely to hit precisely the
same location twice, and thus alleles introduced into
the population via recurrent mutation are very unlikely
to be the same. If neutral mental disorder susceptibility
alleles are specific, in the sense that only one or a few of
all the possible mutations at that locus would affect
mental disorder risk, while all others would not, then
recurrent mutation is too rare an event to replenish lost
susceptibility alleles. In this case, random genetic drift
would lead to loss or fixation of the mental disorder
susceptibility allele. Therefore, models that hypothesize
that mental disorders are complex phenotypes, coded by
specific alleles that are alternatives to the normal alleles,
are not consistent with what is known about the properties
of neutral evolution. However, it is probably more biologically plausible that any mutation along the locus could
increase or decrease mental disorder risk; in this case,
random genetic drift plus recurrent mutation could in
principle account for substantial genetic variation.
The degree of genetic variation contributed by such a
neutral locus, where any mutation affects mental disorder
risk, can be quantified. As already noted, only loci that
are polymorphic (where more than one allele exists in
the population) contribute to genetic variation. Genetic
polymorphism can be measured by H, the proportion of
heterozygotes at a locus in a population. Kimura (1983)
showed that for neutral loci, H ffi 4Nem/(1 þ 4Nem),
where m is the probability of a new mutation at the locus
per individual per generation, and Ne is the effective population size (roughly the harmonic mean of the breeding
population size across generations, which tends to be
close to the minimum actual population size during
genetic bottlenecks). Ne is often estimated to be around
10,000 for humans (Cargill et al. 1999). Assuming m is
around 10 – 6 to 10 – 5 for most loci (Nachman & Crowell
2000), the expected heterozygosity H across neutrally
evolving human loci should be around 4% to 29%.
Because neutral loci have relatively high average values
of H, they can contribute substantially to heritability in
human traits and perhaps mental disorders.
To say that neutral evolution could maintain the genetic
variation underlying mental disorders is very different than
saying that such a process is likely. In sections 4.2 and 4.3,
we review two reasons that neutral evolution is probably
not a general resolution to the paradox of common, heritable, harmful mental disorders, and then we review the
types of phenotypes that might be best explained by a
weaker version of this process (sect. 4.4).
4.2. Ancestral neutrality must be implausibly precise
For an allele to be truly neutral over the evolutionary long
term, the allele must have fitness effects extremely close to
neutrality within each generation. This statement can be
Keller & Miller: Resolving the paradox of heritable mental disorders
quantified simply. For an allele to be neutral (to be governed by genetic drift more than by selection), the selection coefficient s against an allele must be less than 1/
4Ne. Thus, only if the average fitness of people with an
allele is between 99.997% and 100.003% of the fitness of
people without this allele (i.e., if s , 1/40,000) has the
frequency of that allele been governed mostly by neutral
drift across human evolution. This is an extraordinarily
small selection coefficient, equivalent to a difference of
just one offspring more or less than average, not in the
next generation, but 15 generations into the future,
given a roughly constant population size.
Not only must neutral mental disorder susceptibility
alleles have been almost exactly neutral in ancestral
environments, they must have been consistently so. If
the alleles were neutral in most but not all environments,
or in most but not all cultures, or in most but not all bodies,
then averaged across evolutionary time, these alleles
would not be neutral. As we have argued, there are
strong reasons to believe that the mental disorders listed
in Table 2 are fitness-reducing in modern societies. If susceptibility alleles were neutral in ancestral environments
but highly dysfunctional today, this implies very large
G– E interactions. Yet, very large G –E interactions are
implausible, given this consistency requirement that
mental disorder susceptibility alleles had to be unfailingly
neutral across many different ancestral environments.
Although many evolutionary biologists believe that
neutral mutations are the main source of genetic polymorphisms across DNA in general (since most DNA has
no phenotypic effect), few now believe that neutral
mutations are the main source of phenotypically expressed
variation (Ridley 1996). The very fact that neutral alleles
have no fitness effects makes them unlikely to affect
phenotypic development. By contrast, mental disorder
susceptibility alleles do affect the phenotype in modern
environments, and it is likely that they would have done
so in ancestral environments as well. It is hard to believe
that phenotypically expressed alleles associated with conditions that have such harmful effects in modern environments would have been precisely neutral (s , 1/40,000)
across all ancestral environments.
populations (e.g., Burns 2004; T. J. Crow 2000). How
plausible is this? As illustrated in Figure 2, alleles with
even small fitness effects are quickly driven to extinction.
For example, if schizophrenia in Finland has been as disadvantageous over the last 20 generations, as it appears
now (s .50), and is caused by a single recessive allele
with p ¼ .10 (explaining the current disease prevalence
of 1%), it would follow from standard evolutionary
genetics that 42% of Finns were schizophrenic in
1600 – clearly a nonsensical result. Selection on dominant
or additive alleles is even faster. Thus, it is not evolutionarily credible to claim that mental disorders are caused by
one or even a few genes and have a low but significant
prevalence because they became harmful only several
thousand years ago.
4.4. Disorders that ancestral (near) neutrality
might help explain
We have argued that it is highly unlikely that alleles with
substantial fitness-reducing effects today were precisely
and consistently neutral across ancestral environments.
However, alleles affecting certain disorders might have
been much closer to being neutral in ancestral environments, and therefore the modern prevalence rates and
heritabilities of these disorders may be higher than predicted from modern fitness estimates. This is a plausible
hypothesis for heritable disorders that show the hallmarks
of G –E interactions: large cross-cultural variation in
prevalence rates, increased (or decreased) rates in recent
historical time as environments change, and a credible
mismatch between ancestral and modern conditions that
affects the mental disorder.
Data showing that depression rates vary enormously
between cultures, and seem to be rising to very high
levels in industrialized nations (Weissman et al. 1996),
are consistent with – but by no means prove that – G –E
interactions are important in depression. It is also easy
to imagine a credible mismatch scenario for depression.
For example, social support from kin and friends was
probably more available in small-scale ancestral societies
4.3. Ancestral neutrality is hard to reconcile with
modern mental disorder prevalence rates
A strictly neutral hypothesis about mental disorder genetic
risk factors would dictate that all levels of genetic risk have
equal fitness effects. Under such a scenario, any prevalence rate of mental disorders, from 0% to 100%, should
be about equally likely. Contrary to this, Table 1 shows
that the most harmful mental disorders are consistently
rare in an absolute sense, none being more common
than about 2%. If neutral evolution were a general
answer to the paradox, one would have to explain why
the most harmful mental disorders are so consistently
rare. The exceptions that prove the general rule are
late-onset disorders such dementia (which affects about
30% of people over age 75; Thomas et al. 2001), which
are more likely to have been close to selectively neutral
under ancestral conditions.
Perhaps the low frequencies of modern mental disorders suggest that they became fitness-reducing only
recently and are currently being selected out of human
Figure 2. Expected changes (ignoring genetic drift) in allele
frequencies across generations, given different levels of
selection (s) acting on additive deleterious alleles of minor effect.
Keller & Miller: Resolving the paradox of heritable mental disorders
than in modern cities, and such social support may help
rescue people from normal periods of transient depression
(Kessler 1997). Although heritable shyness may have had
little effect on social support in ancestral conditions, “shy
alleles” could decrease the social support available when
times get tough in modern cities, becoming susceptibility
alleles for depression that show strong G –E (or more
specifically, in this case, G-culture) interactions.
Other disorders that could plausibly be affected by
alleles that were more benign in ancestral environments
include: (a) obesity and diabetes, due to unnaturally consistent and appealing food surpluses; (b) asthma, due to
unnatural levels and types of antigens and pollutants;
and (c) addictions to highly purified, evolutionarily novel
drugs, such as heroin or cocaine (Nesse & Berridge
1997). These disorders are heritable within cultures, but
their frequencies differ enormously between cultures
and environments. They have also probably increased in
frequency in cultures most affected within the last
50 to 100 years (Wright & Hastie 2001), as likely environmental risk factors were increasing. It is also reasonable
to assume that their environmental risk factors were
usually absent in ancestral environments. Finally, in
societies with the environmental risk factors, the frequencies of these disorders are not consistently low (e.g.,
obesity rates are approaching 50% among younger U.S.
Nevertheless, it is unlikely that alleles that increase
mental disorder risk today were precisely and consistently
neutral ancestrally – even those alleles that have become
more harmful only recently. Given that natural selection
purges even slightly harmful alleles (Fig. 2), the persistence of alleles that were only close to, but not precisely,
neutral still requires an explanation. The polygenic
mutation-selection paradigm, reviewed in section 6, provides this explanation.
5. Can balancing selection explain common,
harmful, heritable mental disorders?
The genetic variation underlying mental disorders, far
from being invisible to selection, might have been actually
maintained by selection. For example, mental disorders
which look harmful and dysfunctional, and which show
below-average fitness under some conditions, might
show above-average fitness under other conditions. This
type of selection, known as balancing selection, has been
one of the most popular ideas among evolutionary thinkers
for resolving the paradox of common, harmful, heritable
mental disorders (Allen & Sarich 1988; Barrantes-Vidal
2004; Karlsson 1974; Longley 2001; Mealey 1995;
Stevens & Price 2000a), with some researchers even
implying that balancing selection is the only possible resolution to the paradox (D. R. Wilson 1998). One purpose of
this article is to rebut such claims by showing that there
are at least two other potential resolutions to the
paradox: neutral evolution and mutation-selection
Balancing selection may be popular among Darwinian
psychiatrists in part because it keeps natural selection
front and center as the causal force explaining a trait – a
comfortable position for adaptationists. Balancing selection might also be appealing for social and moral
reasons, because it attributes hidden adaptive benefits to
mental disorders in ways that might reduce their social
stigma. Morality aside, how feasible is it that balancing
selection resolves the paradox?
5.1. Natural selection usually depletes genetic variation
As noted in section 2, selection usually leads to genetic
uniformity and therefore depletes heritability. Certain
evolutionary models, such as those for phobias and
depression (e.g., Keller & Nesse 2005; Watson &
Andrews 2002), posit adaptive functions for capacities
that are universal features of human nature, affected by
universal suites of genes (i.e., little or no genetic variation),
and triggered by adverse situations. These explanations are
potentially useful for understanding environmental variation, but do not explain, nor were they intended to
explain, the genetic variation in phobias and depression.
Other evolutionary models hypothesize that heritable
disorders themselves are adaptive without explaining
why the disorders have not fixated in the population.
Consider three recent hidden-benefit models: Guisinger
(2003) viewed symptoms of anorexia as an adaptive
response to fleeing famine under ancestral conditions of
starvation; Sherman (2001) viewed bipolar disorder as an
adaptation to long, severe winters and short summers;
and T. J. Crow (2000) viewed schizophrenia as an inevitable risk arising as a side effect of language evolution (see
also Burns 2004). None of these offer a compelling
explanation for the persistence of heritability in these disorders. If anorexia was simply adaptive under starvation
conditions, then the adaptive anorexia alleles would be
virtually fixed within those human groups whose ancestors
gained such advantages, and the condition should not be
heritable within these groups. In truth, however, very
few people show these symptoms, and the phenotypic
differences between those who do versus those who do
not are largely due to genetic differences (Guisinger
2003). Similar arguments can be made for bipolar disorder
or schizophrenia. These hidden-benefit models may or
may not help explain why humans in general are susceptible to anorexia, bipolar disorder, or schizophrenia, but
they do not explain the central paradox addressed in this
article: why mental disorder susceptibility alleles have
not either fixated, if adaptive, or gone extinct, if maladaptive. This is one of our key points: Explaining heritable
polymorphisms requires special and stringent types of
evolutionary explanations that are different from those
used to explain species-typical traits. Most types of selection offer no explanation for mental disorder heritability.
Balancing selection can.
5.2. Balancing selection is the only type of selection
that actively maintains genetic variation
Balancing selection actively maintains two or more
alternative alleles because their net fitness effects
balance each other out, being positive in certain genetic
or environmental contexts and negative in others. For balancing selection to maintain a stable genetic polymorphism across evolutionary time, (a) the fitness effects of
the alternative alleles must be equal across ancestrally
relevant genetic and environmental contexts, and (b)
some mechanism must assure that these equally fit
Keller & Miller: Resolving the paradox of heritable mental disorders
alleles are not lost by chance (genetic drift). For the most
robust types of balancing selection, if an allele drifts by
chance to a lower level, its fitness increases, which then
buoys its frequency back up. So long as the equilibrium
frequency of one of the alleles is not too low, such a
homeostatic mechanism greatly reduces the risk of
equally fit alleles being lost by genetic drift.
Before assessing the general utility of balancing selection in explaining mental disorders, we review the explanatory power of four specific forms of it: spatial and temporal
variation in selection, heterozygote advantage (also known
as heterosis or overdominance), antagonistic pleiotropy,
and frequency-dependent selection. Although these are
often considered separate evolutionary processes, they
have important common features at the evolutionary
genetic level that give them similar strengths and weaknesses in explaining mental disorders.
Balancing selection can occur when an allele’s fitness oscillates over evolutionary time or location. We are aware of
no models that try to explain mental disorder heritability
by using this mechanism. For this to explain the paradox,
a convincing case would need to be made that mental disorders or their susceptibility alleles were advantageous
across about half of ancestral populations in different
locations or about half of the time, but this seems a
priori unlikely, though not disproved, in light of the consistent harmfulness of mental disorders in current
environments. A deeper, theoretical problem for this
explanation is that no homeostatic mechanism protects
alleles against loss through drift; rather, the evolutionary
oscillations in an allele’s fitness must occur at just the
right rate across time or space to keep the allele from fixating or going extinct (Bürger 2000). Such loss of alleles
would be especially likely in small prehistoric human
Although this mechanism seems theoretically unlikely
to maintain mental disorder susceptibility alleles at equilibrium, it is important to remember that we are catching
but a snapshot of evolution. It is certainly possible that
some susceptibility alleles are at intermediate frequencies
because they are sweeping toward fixation or extinction.
Such a process may be occurring with a susceptibility
allele for heart disease and Alzheimer’s disease:
APOE 4. APOE 4 is the ancestral allele, being rarest
among human groups that have had the longest exposure
to agriculture, and is probably headed over the next
several thousands of years toward extinction (for two
views on why this might be, see Corbo & Scacchi [1999]
and Finch & Sapolsky [1999]). Nevertheless, it is unlikely
that enough alleles are rising or lowering in frequencies for
this to be a general answer to the paradox, given the short
time that alleles with fitness effects are at intermediate
frequencies (Fig. 2).
common allele (AA) at the b-hemoglobin locus are susceptible to malaria, whereas those homozygous for the less
common allele (aa) are more likely to die from sicklecell anemia. However, heterozygotes (Aa) have the best
of both worlds: they do not develop anemia, and they
are much more likely to survive a malarial infection. In
equatorial areas of Africa and Asia where malaria is
endemic, heterozygotes have higher fitness than either
homozygote. If genotypes rather than genes could be
passed to offspring, Aa genotypes would have fixated
long ago, but this cannot happen. For example, matings
between two most-fit heterozygotes nevertheless
produce 14 aa and 14 AA offspring on average. The population frequencies of the two alleles become stable when
the average fitness effects of alleles a and allele A are
equal. Here, a homeostatic mechanism keeps alleles
from being lost through genetic drift: if the frequency of
one allele in the population drifts to a lower level, that
allele has an increased chance of finding itself in a heterozygote body, and its average fitness, and hence frequency,
In the case of sickle-cell anemia, Allison (1954) showed
that, given the fitness estimates for each genotype at the
b-hemoglobin locus, evolutionary genetic theory predicted very well the observed phenotypic frequencies.
The sickle-cell story had a large impact on evolutionary
biologists in the 1950s, and many suggested that heterozygote advantage might be a general explanation for
observed levels of genetic variation in nature (e.g.,
Lerner 1954). More recently, several evolutionists have
theorized that mental disorders such as schizophrenia
(Huxley et al. 1964), bipolar disorder (D. R. Wilson
1998), and depression (D. R. Wilson 2001) are maintained
by heterozygote advantage.
However, for several reasons, evolutionary biologists
have become less enthusiastic about heterozygote advantage as an explanation for persistent heritability in most
traits. First, heterozygote advantage appears to be rare in
nature: Thirty years of intensive research following the
sickle-cell story yielded only six additional examples of
polymorphisms maintained in this way (Endler 1986).
Second, there are theoretical reasons to doubt that
species could sustain widespread maladaptive polymorphisms in this way without going extinct (Crow & Kimura
1970). Third, selection would strongly favor genetic
events that overcome the costs of producing homozygotes,
such as unequal crossover events that positions both A and
a on the same chromosomal arm, so they can be passed on
together without disruption (Ridley 1996), or mutations
that reduce the fitness costs of either homozygote. Such
genetic events become quite likely across a whole
population over evolutionary time, so heterozygote
advantage is likely to be an evolutionarily transient
stopgap. This is consistent with the fact that the a allele
at the b-hemoglobin locus evolved fairly recently
(Hamblin et al. 2002).
5.4. Heterozygote advantage
5.5. Antagonistic pleiotropy
A genetic polymorphism may be maintained when the
heterozygote at some locus has higher fitness than either
homozygote (e.g., genotype Aa has higher fitness than
both AA and aa). The classic example is sickle-cell
anemia. Individuals who are homozygous for the more
Pleiotropy occurs whenever one allele affects more than
one trait. Given that traits do not rely on mutually exclusive sets of genes, pleiotropy is ubiquitous. Antagonistic
pleiotropy, which is also probably ubiquitous, occurs
whenever an allele increases the fitness payoffs of one
5.3. Temporal or spatial variability in fitness landscapes
Keller & Miller: Resolving the paradox of heritable mental disorders
trait but reduces the fitness payoffs of another trait. For
example, an allele might increase fertility but decrease
longevity, or increase intelligence but decrease emotional
Generally, this process leads to the fixation of whichever
allele has the highest fitness, averaged across the various
effects it has on different traits. Even if the net fitness
effects of two alternative alleles are precisely equal,
which is a priori unlikely, there is no homeostatic mechanism that counteracts the homogenizing effect of genetic
drift (Curtisinger et al. 1994; Hedrick 1999; Prout 1999).
In fact, this theoretical work suggests that antagonistic
pleiotropy is likely to maintain genetic polymorphisms
only under a highly restrictive scenario: when individuals
with both alleles receive the fitness benefits but not the
costs from each allele – a situation called reversal of
dominance. In this situation, heterozygotes would have
the highest fitness, a scenario conceptually equivalent to
heterozygote advantage, and which shares the same explanatory weaknesses. The conclusion from theoreticians is
that antagonistic pleiotropy cannot maintain genetic
variation on its own; it requires a very special type of
allelic effect, reversal of dominance, which evolutionary
biologists consider unlikely.
Despite these theoretical concerns, antagonistic
pleiotropy is probably the most common evolutionary
explanation for the persistence of susceptibility alleles.
Several researchers have hypothesized that susceptibility
alleles underlying bipolar disorder and schizophrenia
have two effects: one, to increase creativity, but the
second, to increase the risk for the mental disorder
(Barrantes-Vidal 2004; Karlsson 1974). These susceptibility alleles are thought to persist because their negative
fitness effects from mental disorder risk are precisely
offset by their benefits from creativity. The idea that
mental disorders are associated with higher creativity is
widespread, and supported by some biographical evidence
(Jamison 1993) and evidence that relatives of those with
mental disorders have higher creativity (reviewed by
O’Reilly et al. 2001). However, a literature review of 29
studies found little support for the idea that highly creative
people showed an increased rate of mental disorders
(Waddell 1998).
5.6. Frequency-dependent selection
Frequency-dependent selection (or more technically,
negative frequency-dependent selection) occurs when
alleles’ fitness effects increase as they become rarer. This
process can maintain a stable mix of alleles resulting in
persistent trait heritability. Heterozygote advantage can
be seen as a special case of this process. Frequencydependent selection more generally occurs when individuals compete for different resources, such that individuals
who are rare relative to their preferred resource are
favored (Barton & Keightley 2002).
The classic example of frequency dependence is the
evolutionary maintenance of the 50:50 sex ratio (Fisher
1930/1999). If males outnumber females, females necessarily have higher average reproductive success than do
males. A mutation increasing the probability of having
daughters would be positively selected, and would
spread in the population until females began to outnumber males, in which case selection would begin to favor
having sons. The evolutionary equilibrium is that both
strategies (being male or being female) reach equal
frequency, although, in other cases, alternative strategies
may have non-equal equilibrium frequencies. Frequencydependent selection can maintain high levels of heritable
genetic variation for as long as the selection pressures
For a few mental disorders such as psychopathy, frequency dependence may be a plausible model. Mealey
(1995) argued, forcefully in our opinion, that psychopathy
persists at a low base rate as a socially parasitic strategy: it
brings high fitness benefits when rare, but becomes less
rewarding at higher frequencies because of increased
anti-cheater vigilance in the population. Indeed, at the
current low base rate (around 1%), male psychopaths
seem to have higher-than-average fitness, at least in
modern environments – unlike almost all other mental
disorders listed in Table 1. In general, frequencydependent selection can explain polymorphic alleles only
when there is a credible explanation of why each allele’s
fitness increases as its frequency decreases. This is a fairly
high standard of evidence. Moreover, there are several problems with balancing selection in general as an explanation
for mental disorders, which we explore next.
5.7. General problems with balancing selection
explanations for mental disorder susceptibility alleles
Mental disorders are not a random sample of human traits;
they are considered “disorders” precisely because they
have salient maladaptive outcomes. Rare phenotypes
with such severe costs, as opposed to common phenotypes
that are not debilitating, are probably the least likely
candidates for traits maintained by balancing selection.
This is because the devastating negative effects of
susceptibility alleles must be balanced by commensurately
large, and therefore probably noticeable, positive effects
(e.g., sickle-cell anemia being balanced out by malarial
resistance). Balancing selection may explain some heritable personality traits such as extroversion and some
personality disorders such as psychopathy. Yet it seems a
poor candidate as a general explanation of the susceptibility alleles of mental disorders, since their susceptibility
alleles would have to show some hidden adaptive benefits
that counteract the strongly maladaptive symptoms of
these mental disorders.
Another problem for models of both spatio-temporal
variation and frequency-dependent selection is that behavioral flexibility, as opposed to fixed, heritable strategies,
would probably be favored in the face of differing fitness
landscapes (Tooby & Cosmides 1990; although see D. S.
Wilson 1994). Fixed, heritable strategies make sense for
basic morphological specializations such as growing a
male or female body, when it is hard to switch from one
to the other after growth. However, the whole point of
growing a central nervous system is that different behavioral strategies, which have context-dependent fitness
payoffs, can be pursued by the same individual across
different situations. Such flexibility circumvents the costs
of pursuing fixed strategies when their frequencies are at
the wrong level to maximize fitness. Given the extraordinary behavioral flexibility of the human brain, it would
be puzzling if such genetically fixed strategies explained
mental disorder heritability. Despite these two broad
Keller & Miller: Resolving the paradox of heritable mental disorders
problems, and the problems specific to the various types of
balancing selection, balancing selection cannot be ruled
out as a resolution to the paradox on purely theoretical
grounds. In section 7, we review several pieces of empirical evidence that support our expectations that balancing
selection is not a general explanation for mental disorder
susceptibility alleles.
alleles, this mechanism still begs for an explanation of
why near-neutral alleles have not fixated or gone
extinct – a topic we turn to next.
5.8. What balancing selection might explain
The simplest polygenic mutation-selection model elegantly
parallels the single-gene models described earlier: the
equilibrium genetic variation (VG) maintained in a trait
affected by many loci is VG ¼ VM/s, where VM is the
increase in a trait’s genetic variation due to new, harmful
mutations per generation, and s is the average selection
coefficient against these mutations (Barton 1990). It generally takes a while for these harmful mutations to work their
way out of the gene pool. For example, a mutation causing
a 1% reduction in fitness will persist in the population until
it has passed through an average of 100 individuals
(Garcı́a-Dorado et al. 2003). Mutations with the most
harmful effects are removed the fastest, so if one is
observed, it is probably rare and of recent origin.
Mutations with milder effects are removed more slowly,
so they tend to be more common (although still very
uncommon in an absolute sense) and older, inherited
from parents, grandparents, and so forth. Therefore,
genetic variation caused by mutation-selection balance is
predominantly the result of old mutations that have yet
to go extinct, rather than new mutations, a point that is
commonly misunderstood. Most mutations are a family
legacy, not an individual foible.
Several recent theoretical papers have emphasized the
role of mutation in maladaptive human traits (Gangestad &
Yeo 1997; Hughes & Burleson 2000) and late-onset
diseases (Wright et al. 2003). Such polygenic mutationselection models suggest that much of the persistent
heritability in traits may be due to a large number of
harmful alleles that are individually very rare at any
given locus in the population, but that are collectively
very common across loci. These models recognize that
we do not live in the best of all possible worlds. Genetic
information is constantly and inevitably eroded by
genetic copying mistakes: mutations. Applied to human
mental disorders, mutation-selection models suggest
that, if a mental disorder appears maladaptive, maybe it
really is maladaptive – and always has been.
Balancing selection might, in theory, maintain mental
disorder susceptibility alleles for reasons completely
unrelated to mental disorder symptoms. For example,
pathogens and parasites are usually poorly adapted to
attacking the rarest host genotypes, so rare alleles may
help protect the host (Garrigan & Hedrick 2003;
Haldane 1949). This anti-pathogen variation could give
rise to mental trait variation as a side effect (Tooby &
Cosmides 1990), and some researchers have suggested
this might explain the high prevalence of schizophrenia
susceptibility alleles (J. S. Brown 2003). For this to work,
the fitness benefits of improved host defense must
outweigh the fitness costs of increased mental disorder
risk (Turelli & Barton 2004). Because most loci probably
do not affect immunological systems, the vast majority of
loci are probably unaffected by parasite-host coevolution.
Moreover, selection should have favored minimal
overlap between the genes that control anti-pathogen
defenses and those that affect other systems, such as the
nervous system, although there may be a limit in how far
natural selection can go in removing such pleiotropic
effects of genes. Although it is certainly possible that
some psychological variation is a by-product of frequencydependent selection for other traits, empirical evidence
discussed in section 7 makes it unlikely to be a general
resolution to the paradox.
We have argued on both theoretical and empirical
grounds that antagonistic pleiotropy is unlikely to explain
the persistence of mental disorder susceptibility alleles,
but a weaker version of it may work better. This version
suggests that alleles with conflicting fitness effects on
different traits should tend to be closer to neutral than
alleles without such antagonistic effects, so perhaps they
will persist longer at intermediate frequencies and contribute more to heritable variation. If such a near-neutral
allele has opposite fitness effects on two traits, those
traits should show a negative genetic correlation (Lande
1982). More generally, if antagonistic pleiotropy accounts
for substantial genetic variation, most genetic correlations
between fitness-related traits should be negative. This
logic is compelling, but the evidence among animal traits
is not very supportive. A meta-analysis of genetic correlations between fitness-related traits in nonhumans
found that 61% were positive (i.e., the higher fitness end
of one trait tended to go with higher fitness end of other
traits; Roff 1997) – a result less congruent with antagonistic pleiotropy than with polygenic mutation-selection
balance models (Charlesworth 1990). Nevertheless, it is
plausible that some portion of the genetic variation underlying mental disorders is due to near-neutral alleles that
increase mental disorder risk under certain genetic or
environmental conditions, but that have some positive
benefits in other conditions. Much like the scenario discussed in section 4 for non-pleiotropic near-neutral
6. Can polygenic mutation-selection balance
explain common, harmful, heritable mental
6.1. Is polygenic mutation-selection a viable explanation
for the genetic variation in traits?
For several reasons that now appear misguided, researchers have often doubted that mutation-selection could
explain mental disorder susceptibility alleles. First, the
results of many animal studies seemed to suggest that
just a few loci (around 2– 20) account for much of the genetic variation in traits that had been studied (Falconer &
Mackay 1996) – too few for mutation-selection balance
to play much of a role in mental disorders. However,
there are good reasons to think these studies underestimated
the number of loci and overestimated their effect sizes
(Barton & Keightley 2002). Moreover, the traits analyzed in
these studies generally have little relevance to fitness (e.g.,
number of abdominal bristles in fruit flies), and mutation
Keller & Miller: Resolving the paradox of heritable mental disorders
plus drift at few loci can maintain substantial genetic
variation in nearly neutral traits. Second, it is estimated
that approximately 7 million single-nucleotide polymorphisms (SNPs) have minor allele frequencies greater than 5%
(Kruglyak & Nickerson 2001). However, more than 98% of
these 7 million SNPs are outside of protein coding regions
and are unlikely to affect mental disorder risk (Wright et al.
2003). SNPs that do affect protein production tend to have
minor allele frequencies below 5% (Fay et al. 2001), which
is consistent with mutation-selection balance.
Third, as discussed earlier (sects. 2, 4.2, and 5.1), there
were strong theoretical expectations that maladaptive
states should be rare in nature. Fisher’s Fundamental
Theorem seemed to suggest that additive genetic variation
should be lowest in traits under the strongest selection.
This prediction seems supported by observations that
traits under more intense selection have lower heritability
estimates (Roff & Mousseau 1987). However, heritability
is but one way to measure additive genetic variation, and
alternative measures of additive genetic variation has
turned the canonical story about genetic variation in
fitness related traits on its head.
Heritability (h 2 ¼ VA/VP) is the proportion of total phenotypic variation (VP) due to additive genetic effects (VA).
VP is influenced by all sources of variation – not just VA,
but also by environmental variation, random noise, and
non-additive genetic effects. Low heritability may well be
a result of low VA, but it might also be caused by high VP
(Charlesworth 1987; Price & Schluter 1991). Charlesworth
(1984), Houle (1992), and others have argued that
coefficient of additive genetic variation (CVA ¼ VA =x)
is a better way to measure VA (i.e., to remove scale dependence of VA), because it is standardized by the trait’s mean
(x) rather than by VP and is therefore not confounded by
environmental factors, random noise, or nonadditive
genetic effects. (Unfortunately, the use of CVA requires
that the trait can be measured on ratio scales, such as
number or time, and is therefore unsuitable for measuring
genetic variation in most psychological traits.)
In a seminal study, Houle (1992) found that traits
under stronger selection show substantially higher meanstandardized additive genetic variation than do traits
under weaker selection, despite showing lower heritability.
For example, fruit-fly wing length (a trait under relatively
weak selection) has a heritability of .36, whereas number
of offspring (a trait under intense selection) has a heritability of only .06. However, wing length has a CVA of only 1.6,
whereas number of offspring has a CVA of 11.9. Across
many such comparisons, the mean-standardized VA of
traits under the strongest selection is three to ten times
higher than that for traits under weaker selection, the
opposite of what Fisher’s Fundamental Theorem would
seem to predict. Similar results have been replicated
now in many species, including humans (Hughes &
Burleson 2000). These results were astonishing at first
and created quite a stir among evolutionary geneticists,
leading to a paradox that both parallels and informs the
paradox of common, harmful, heritable mental disorders.
6.2. The watershed model explains why traits under the
strongest selection have the highest genetic variation
Traits under the most intense selection ( fitness-related
traits, such as successful growth or mating) tend to
require the adaptive functioning of many subsidiary
biological and behavioral processes, and so depend on
very many genes (Charlesworth 1987; Houle 1992; Price &
Schluter 1991). The most massively polygenic “trait” is,
of course, fitness itself – successful survival and reproduction – which requires the functional coordination of every
adaptive mechanism in the body. The mutational “target
size” of fitness is quite obviously the entire genome with
any effect on fitness, which is probably the vast majority
of genes with any phenotypic effect.
The biological network of mechanisms that must function together to create adaptive behaviors can be conceptualized by using a watershed analogy. Much like the
numerous tributaries of the Amazonian watershed that
coalesce and eventually empty into the Atlantic Ocean,
there are many “upstream” micro-biological processes
(e.g., rates of neuron proliferation, dendritic pruning,
glucose metabolism) that flow into (affect) further “downstream” macro-biological processes (e.g., finding food,
making friends, securing mates). A mutation at a locus
that affects an upstream process disrupts not only that
upstream process, but also every trait downstream of
that process. A slightly harmful mutation that affects dendritic pruning may not affect glucose metabolism, but will
probably undermine downstream processes such as learning ability, attracting mates, and eventually fitness itself.
Figure 3 illustrates this watershed analogy.
The watershed analogy suggests that fitness-related
traits have high additive genetic variation because they
integrate many processes, and so are massively polygenic.
Thus, they are vulnerable to harmful mutations at many
loci, and have higher additive genetic variation due to
new and old mutations. Fitness-related traits have high
VA, despite being under intense selection, not because of
it. Their high VA reflects that they tend to be massively
Figure 3. The watershed model of the pathways connecting
upstream genes to downstream phenotypes. Mutations at
specific loci (1a, 1b) disrupt narrowly defined mechanisms such
as transmission of dopamine in the prefrontal cortex (2b). This
and other narrowly defined mechanisms contribute noise to
more broadly defined mechanisms, such as working memory
(3c). Working memory in conjunction with several other
mechanisms (3a, 3b, 3d) affects observable phenotypes, such as
cognitive ability (4). If enough noise is present in particular
upstream processes, specific behavioral syndromes may arise,
such as mental disorder symptoms. All tributaries eventually
flow into fitness. (Reprinted, with permission, from the Annual
Review of Clinical Psychology, volume 2. # 2006 by Annual
Reviews [Cannon & Keller 2005])
Keller & Miller: Resolving the paradox of heritable mental disorders
polygenic. This is not symmetrical: neutral traits are not
necessarily influenced by few genes, but fitness-related
traits are almost always influenced by many genes. There
is now a good deal of support for this model, at least in
fruit flies – the animal model of choice for evolutionary
geneticists. Among the most compelling pieces of evidence
are the high, positive intercorrelations between (a) the
estimated number of loci influencing different traits,
(b) the estimated trait-level mutation rates, and (c) traits’
CVA (Houle 1998). Charlesworth and Hughes (1999)
further estimated that rare, harmful mutations account
for 33% to 66% of the additive genetic variation in
fitness-related traits in fruit flies.
The watershed model also clarifies why fitness-related
traits typically have very high phenotypic variation and
therefore moderate to low heritabilities. Downstream
traits accumulate any type of noise from upstream traits –
not only mutational noise, but also environmental noise
(e.g., bad luck with injuries, predators, pathogens, and
mates), random noise (e.g., the inherent stochasticity of
development), and non-additive genetic effects. Because
selection has much less power to reduce the variation in
these latter factors compared to additive genetic variation,
these factors tend to be proportionately more influential
for fitness-related traits, leading to their lower heritabilities (Houle 1992; Merilä & Sheldon 1999).
6.3. The mutational target size of the human brain
The watershed model suggests that fitness-related traits
have high genetic variation because they are massively
polygenic. How might the watershed model help explain
mental disorders? The answer depends upon how many
loci influence the mechanisms that, when dysfunctional,
cause the behavioral syndromes defined as mental
Consider the complexity of human brain function in
watershed terms. The human brain is the most complex
system known to science, with about 100 billion neurons
and about a thousand times that many synapses. At least
55% of coding DNA is probably expressed in the human
brain (Sandberg et al. 2000). Thus, the brain has an
enormous mutational target size – out of the 25,000
protein-coding genes estimated in the human genome,
mutations in at least half of them are likely to disrupt
brain function, and hence behavior, to some extent
(Prokosch et al. 2005).
Yet, there is more to the human genome than proteincoding regions. About as much non-coding DNA as
coding DNA is evolutionarily constrained between
species, implying that non-coding, regulatory regions are
about as important to fitness as protein-coding regions
(Keightley & Gaffney 2003). Importantly, non-coding
regulatory regions of DNA rarely contribute to Mendelian
disorders (McKusick 1998). Thus, harmful mutations in
non-coding, regulatory regions seem to have mainly
subtle quantitative effects rather than producing dramatic
Mendelian catastrophes, and may be especially relevant in
explaining the continuously distributed liabilities thought
to underlie mental disorders.
How high is the typical human mutation load in
brain-expressed loci? Based on conservative estimates,
each human carries about 500 to 2,000 slightly harmful
older point mutations inherited from ancestors in
protein-coding regions (Fay et al. 2001; Sunyaev et al.
2001), plus an average of one or two new fitness-reducing
mutations (Eyre-Walker & Keightley 1999). These
mutation-load estimates should be at least doubled to
account for mutations in non-coding, regulatory DNA,
and should be increased slightly to account for mutations
involving insertions, deletions, and other changes to chromosomal structure. Given that perhaps half of these
mutations affect the brain, we estimate that the average
human brain is disrupted by an average of at least 500
genetic mutations.
Apart from a high average mutation load, humans are
likely to show high variation in mutational effects. If the
numbers of mutations across individuals follows a
Poisson distribution, as it would under random mating
(S. Gangestad, personal communication, March 3, 2005),
the mean and variance in numbers of mutations would
be equal, implying
pffiffiffiffiffiffiffiffi a standard deviation of at least 22
mutations ( 500). However, because humans probably
assortatively mate for genetic quality through mutual
mate choice (Miller 2000a), the variation in mutation
number would be further amplified, so some people
should inherit many fewer, and others many more, brainexpressed mutations than average. Moreover, the genetic
variation caused by these varying numbers of mutations
would be higher still, given that different mutations vary
enormously in their effect sizes. The end result will be
continuous distributions with respect to almost all
psychological dimensions. Individuals with a high load of
mutations that affect a particular configuration of upstream
cognitive processes would be at higher risk of having
mental disorders associated with deficits in downstream
behaviors, and would tend to pass this risk on to their
offspring. The importance of brain-expressed mutations
is consistent with evidence for good genes sexual selection
for human mental traits (e.g., Haselton & Miller 2006;
Keller, in press; Miller 2000a; 2000b; 2000c; Miller &
Todd 1998; Prokosch et al. 2005; Shaner et al. 2004).
6.4. How many loci affect mental disorders?
Before considering this question, it is important to note
that the number of loci affecting a trait means something
different to psychiatric geneticists versus evolutionary geneticists. To psychiatric geneticists, this phrase usually
refers only to the loci that currently contribute to the
bulk of a trait’s genetic variation, which we refer to as
the number of polymorphic loci. To evolutionary geneticists, however, the “number of loci affecting a trait”
usually refers to the much larger number of loci that
could affect the trait if those loci were polymorphic. It is
this latter meaning, which we refer to as the number of
potential loci, that is relevant to mutation-selection
models. Pritchard (2001) estimated that only about 10%
of a trait’s potential loci will actually be polymorphic at
any given time (assuming weak selection), a figure corroborated using a different method by Rudan et al. (2003b).
Recent reviews have invariably concluded that polygenic models (including at least two polymorphic loci)
best describe the inheritance of mental disorders such as
unipolar depression (Johansson et al. 2001), bipolar disorder (Blackwood et al. 2001), schizophrenia (Sobell
et al. 2002), mental retardation (Plomin 1999), and
autism (Folstein & Rosen-Sheidley 2001). Beyond this,
Keller & Miller: Resolving the paradox of heritable mental disorders
however, little is known regarding how many polymorphic
loci affect mental disorders, because there has been so
little success in actually finding them or modeling their
numbers. For example, the data on schizophrenia inheritance are fit equally well by models that predict just a
few polymorphic loci (e.g., Risch 1990) and by models
that predict an “infinite” number of loci (e.g., Sullivan
et al. 2003). The differences in conclusions are largely
due to differences in assumptions (additive or epistatic
allelic effects; a distinctive syndrome or an extreme of a
normally distributed liability) about which no definitive
information is available. Nonetheless, it is becoming
clear from gene-mapping studies that many loci, at least
5 –10 and perhaps many more, must influence the beststudied mental disorders: schizophrenia and bipolar disorder (Kendler & Greenspan, in press).
Rather than further considering assumption-laden
models or preliminary empirical results, perhaps it is
worthwhile to take a step back and consider carefully
what mental disorders, as categories, truly are. Mental disorders are much less objective qualities than age, gender,
height, or white blood cell count. Mental disorders are
constellations of aberrant behaviors that were lexicalized
as unitary disorders by psychiatrists in the nineteenth
and early twentieth centuries. There are several possible
reasons why mental disorder categories were chosen as
they were. First, maybe each mental disorder really has
a unitary etiology – a single consistent genetic, neurological, or environmental cause – but few psychiatrists
subscribe to such a notion today. Most mental disorders
show too much heterogeneity within categories, comorbidity across categories, and continuity with normality, to
qualify as discrete, unitary diseases.
Second, as Bleuler (1911) and Jaspers (1923) argued
regarding “the schizophrenias,” an apparently unitary
mental disorder may be a heterogeneous group of dysfunctions in different mechanisms whose final common behavioral pathways lead to similar symptoms. Upstream
biological processes that ultimately affect abstract psychological traits are largely hidden from human perception
(see 3a and 3b in Fig. 3). They are microscopic neuroanatomical problems hidden within the brain. When these
upstream processes dysfunction, humans can usually
observe only the downstream behavioral outcomes, and
not the specific dysfunctions themselves. Such etiological
heterogeneity becomes apparent only in rare cases when
dysfunctions in specific upstream mechanisms leave a
unique phenotypic signature, in addition to normal
symptoms of mental disorders. For example, at least 20
genetic conditions, such as the XXX and XYY karyotypes,
congenital adrenal hyperplasia, Wilson’s disease, and velocardiofacial syndrome, increase schizophrenic symptoms
(Propping 1983). As Vogel and Motulsky (1997) put it,
Survival of this diagnostic concept was achieved – at least in
part – by an interesting strategy: whenever symptoms characteristic of schizophrenia were observed in association with
findings that suggested organic disease, the diagnosis of schizophrenia was withheld. . . . [W]hen all [such] patients . . . were
excluded, a disease group remained for which specific causative factors could not be found.” (p. 700)
Third, and most radically, a mental disorder may be perceived as a coherent category not because it is a “natural
kind” with a common etiology at any level, but because
it was evolutionarily or culturally adaptive for people to
categorize others in particular ways in order to make
certain social decisions about them. Thus, insanity may
be like ugliness, dishonesty, or aggressiveness – things to
avoid and stigmatize in social and sexual interactions – not
because they have a unitary etiology, but because they
have a common set of fitness costs for observers.
The latter two explanations are not mutually exclusive,
of course. We find it likely that apparently unitary
mental disorders are partly in the dysfunctions of the sufferer, and partly in the person-perception adaptations of
the beholder. Mental disorder categories may reflect a
mix of historical convention, diagnostic convenience,
innate categorization biases in person perception, and
common final pathways of partially overlapping yet distinct
dysfunctions. This suggests that the number of loci affecting a mental disorder depends in large part on the way
human minds categorize behavioral symptoms. The
search for endophenotypes (Cannon & Keller 2005;
Gottesman & Gould 2003) is critically important because
it enables researchers to discern more directly the varied
upstream processes whose dysfunctions increase mental
disorder risk, while relying less on perceived symptom
similarity. The most useful endophenotypes should be
those that are further upstream and etiologically less
complex. If the past is any guide, the heterogeneity documented in mental disorders so far may be only the tip of
the iceberg. Underneath a few simplistic mental disorder
categories may lie a vast diversity of potential behaviorimpairing mutations across the thousands of genes
involved in brain development.
7. Empirical evidence on the three models for
common, harmful, heritable mental disorders
We have reviewed several theoretical reasons why polygenic mutation-selection balance may explain the genetic
variation underlying mental disorders, much as it explains
rare Mendelian disorders. We have also presented some
theoretical and empirical reasons to doubt that neutral
evolution or balancing selection are good general resolutions to the paradox, although they may play a role
under certain specific conditions that we delineated. Fortunately, empirical evidence can help distinguish which of
these models goes the farthest in explaining mental disorder susceptibility alleles. We now review six lines of evidence that, taken together, strongly suggest that harmful
mutations underlie a substantial portion of the genetic
risk in mental disorders.
7.1. Fitness and mental disorders
As noted above (sect. 3.3), mental disorders are associated
with lower fertility (due in large part to reduced mating
opportunities; see Table 2) and a high level of disability
in modern industrialized environments. This is consistent
with mutation-selection models, but is less easily reconciled with models of ancestral neutrality and balancing
selection. There is one classic example of balancing
selection maintaining a highly deleterious condition in
humans – sickle-cell anemia – where the strong selection
against anemia is balanced by strong selection favoring
malarial resistance. To our knowledge, such benefits that
balance the harm done by mental disorders have not
Keller & Miller: Resolving the paradox of heritable mental disorders
been reliably documented for any mental disorder.
Indeed, recent evidence on schizophrenia casts doubt
that susceptibility alleles for schizophrenia have any
hidden benefits, at least in modern environments. If
schizophrenia susceptibility alleles are being maintained
by either heterozygote advantage or antagonistic pleiotropy, non-affected siblings of schizophrenics should
have higher fitness than the general population.
However, the best-controlled and largest study of its
kind found that 24,000 siblings of 11,000 schizophrenics
(sample sizes from all previous studies were fewer than
200 schizophrenics) had the same reproductive success
(99.8%) compared to the general population (Haukka
et al. 2003). The 2003 study by Haukka and colleagues
had plenty of power to detect even minor differences in
fitness among relatives of schizophrenics, such as those
(around 5%) that might be required if heterozygote
advantage maintains the susceptibility allele (Allen &
Sarich 1988). Because modern reproductive success may
not correlate with ancestral fitness, as we discussed
earlier (sects. 3.3 and 4), such evidence does not disprove
heterozygote advantage or antagonistic pleiotropy as
mechanisms responsible for schizophrenia, but it does
weigh against them.
7.2. The effect of trauma on mental disorders
Major genetic abnormalities and environmental insults
tend to increase rather than decrease mental disorder
risk. For example, chromosomal abnormalities such as
trisomy, translocations, and mutations of major effect
cause syndromes consistent with autism, mental retardation, schizophrenia, bipolar disorder, and major
depression (reviewed in MacIntyre et al. 2003). Traumatic
brain injuries increase the risk of mental retardation,
schizophrenia, anxiety disorders, and depression (Max et al.
1998; Rao & Lyketsos 2000; Schoenhuber & Gentilini
1988). This type of evidence poses a serious challenge to
balancing selection models, particularly those that posit
that mental disorders themselves are alternative, complex
adaptations maintained by selection. Given that adaptations require the complex coordination of many mechanisms, traumas should disrupt adaptive complexity, not lead
to it. Receiving a blow to the head, for example, should not
lead to higher intelligence or attractiveness. The direction
in which traits move after traumas provide information
about the direction of fitness. The mutation-selection
model seems most consistent with this evidence: the fact
that major phenotypic disruptions (traumas and genetic
abnormalities) increase the risk for mental disorders is
consistent with the hypothesis that minor phenotypic disruptions (mutations of generally minor effect) do likewise.
7.3. The effect of paternal age on mental disorders
Female humans are born with their full supply of 400þ
eggs, and these eggs have gone through only 23 replications, a number that does not change as females age. By
contrast, males must continue to produce new sperm
throughout life. At age 15, sperm cells have gone
through about 35 chromosomal replications, increasing
to 380 by age 30, and 840 by age 50 (J. F. Crow 2000).
Because each chromosomal replication carries a small
chance of a copying error (mutation), the probability of
germ-line mutations increases, at a greater than linear
rate, with paternal age. Consistent with a mutationselection model, higher paternal age, but not maternal
age, is associated not only with many Mendelian disorders,
but also – tellingly – with lower intelligence (Auroux et al.
1989), and an increased risk of mental retardation (Zhang
1992), schizophrenia (Brown et al. 2002; Malaspina et al.
2001; Sipos et al. 2004; although see Pulver et al. 2004),
and mental disorders in general (Hare & Moran 1979).
Perhaps 15% to 25% of all cases of schizophrenia are a
result of this paternal age effect (Malaspina et al. 2001;
Sipos et al. 2004), which would be consistent with most
other cases being a result of milder, older, more numerous
mutations. These paternal age effects are a direct challenge to neutral and balancing selection explanations of
mental disorders, but are exactly what would be expected
under a mutation-selection model (J. F. Crow 2000).
7.4. The effect of inbreeding on mental disorders
Older harmful mutations tend to be more recessive than
new mutations because selection quickly removes
mutations with the largest and most dominant harmful
effects. Inbreeding, or mating between close genetic relatives, reveals the full harmful effects of these old, mostly
recessive mutations because offspring of close relatives
are homozygous at more loci. Consistent with a mutational
role in mental disorder risk, inbreeding in humans has
been associated with mental retardation and low
intelligence (Vogel & Motulsky 1997), unipolar and
bipolar depression (Rudan et al. 2003a), and schizophrenia
(Abaskuliev & Skoblo 1975; Bulayeva et al. 2005; Gindilis
et al. 1989; Rudan et al. 2003a; although see Chaleby &
Tuma 1986; Saugstad & Ödegard 1986). If true, this
phenomenon of inbreeding depression not only implicates
partially recessive harmful mutations in mental disorder
risk among non-inbred populations; it also shows that
selection acted to minimize mental disorder risk in the
ancestral past. It is well known in evolutionary genetics
that inbreeding depression occurs among traits that have
been under directional selection. Ancestral neutrality
and balancing selection cannot explain why inbreeding
increases mental disorder rates. For example, if
schizophrenia risk alleles were maintained by frequency
dependence, then inbreeding would be as likely to
reduce as to increase schizophrenia risk. Selection only
enriches the gene pool with recessive alleles when higher
trait values (in this case, higher mental disorder risk) lead
to lower fitness.
7.5. Comorbidity between mental disorders
Studies have typically found strong associations between
mental disorders; for example, a recent study found that
mental disorder comorbidity ranged from 44% to 94%,
depending on the mental disorder (Jacobi et al. 2004).
This comorbidity appears to be driven in part by pleiotropic genes that simultaneously affect different disorders:
there are positive genetic correlations between unipolar
depression and generalized anxiety disorder (Kendler
et al. 1992), unipolar depression and bipolar disorder
(McGuffin et al. 2003), bipolar disorder and schizophrenia
(Craddock et al. 2005), autism and unipolar depression
(Piven & Palmer 1999), and schizophrenia and several
Keller & Miller: Resolving the paradox of heritable mental disorders
types of mental retardation (Vogel & Motulsky 1997).
Mental disorders are also highly comorbid with many heritable somatic conditions, such as asthma and hypertension
(Buist-Bouwman et al. 2005). Comorbidity and positive
genetic correlations among mental disorders are nicely
explained by mutation-selection models, but would not
be expected under ancestral neutrality or balancing selection models. For example, if susceptibility alleles for
schizophrenia and bipolar disorder were both ancestrally
neutral in their fitness effects, or if their alleles were maintained by balancing selection, there would no particular
reason for them to become genetically correlated with
each other. On the other hand, if mental disorders are
influenced by mutations at hundreds of (potential) loci,
which is in the neighborhood of what would be needed
for mutation-selection models to explain their prevalence,
it would be vanishingly unlikely for each disorder to arise
through a mutually exclusive set of genes, given that the
human genome includes only about 25,000 proteincoding loci. The genetic risk alleles for mental disorders
must overlap quite a lot. This is where the watershed
metaphor falls apart: a small mutation (a tributary) can
contribute to many different symptoms (rivers); the
mapping from genes to mental disorders is many-tomany rather than many-to-one.
7.6. The likely frequencies of mental disorder
susceptibility alleles
To guide the search for mental disorder susceptibility
alleles, it is crucial to know whether susceptibility alleles
are common (one or a few susceptibility alleles per
disease locus at high frequencies in the population), or
individually rare (one exceedingly predominant nonsusceptibility wild-type allele and many different rare
susceptibility alleles at each disease locus). Gene
mappers differentiate these two possibilities; the first is
called the common disease, common variant (CDCV)
hypothesis, whereas the latter has been dubbed the
common disease, rare variant (CDRV) hypothesis
(Wright et al. 2003). To the degree that the CDCV hypothesis reflects the state of the world, current methods of gene
mapping should suffice for finding mental disorder susceptibility alleles. On the other hand, the CDRV model
suggests that future progress in locating susceptibility
alleles will continue to be slow, because the statistical
association, between common “marker” alleles and rare
susceptibility alleles that gene mapping requires, will be
low or nonexistent (Terwilliger & Weiss 1998; Weiss &
Clark 2002; Wright & Hastie 2001). Moreover, if susceptibility alleles are rare, they must exist at a large number of
loci to explain mental disorder rates and heritabilities,
which would further decrease the power of genemapping studies. Understandably, the CDRV model has
not been well received among psychiatric geneticists. A
speaker at a major gene-mapping conference conceded
that this CDRV scenario was too depressing to contemplate, and so it was better to proceed as if it were not
true (Wright & Hastie 2001).
The three models of selection each leave different signatures in the genome that correspond roughly to the CDCV
model or the CDRV model (Bamshad & Wooding 2003;
Kreitman 2000). One of the strongest predictions from
practically every model of balancing selection is that it
will lead to relatively few polymorphic loci, each harboring
just a few (usually two) different alleles at fairly high frequencies (minor allele frequencies greater than about
5%, which we call common alleles), that account for most
of the genetic variation in the trait (Barton & Keightley
2002; Roff 1997). This appears to hold whether the balancing selection is for discrete or continuous trait variation
(Mani et al. 1990).
Whereas the prediction that balancing selection leads to
common alleles appears robust, the prediction that balancing selection leads to just one or a few loci being polymorphic is more nuanced. The latter prediction applies
only to the number of loci that influence traits directly
under balancing selection. If a trait is not under balancing
selection (i.e., is under directional or stabilizing selection),
some of the alleles that influence the trait may nevertheless be pleiotropic and under balancing selection for
reasons unrelated to the trait in question (e.g., Turelli &
Barton 2004). In this case, there is no limit on the
number of loci under balancing selection that might
influence the trait. For example, it is possible that schizophrenia risk has always been maladaptive (under
directional selection), but that many of the (pleiotropic)
loci affecting schizophrenia risk also affect immune
functioning and have been under frequency-dependent
selection for immunity (see sect. 5.8). Therefore, if mental
disorder risk is a pleiotropic side effect of genes that are
under balancing selection on other traits, then common
alleles – but at an unknown number of loci – should be
responsible for most of the mental disorder genetic
risk. If mental disorder risk is directly under balancing
selection, as many Darwinian psychiatrists have
postulated, common alleles at just a few loci should be
responsible for most of the genetic risk of mental
disorders. Regardless, if balancing selection maintains
susceptibility alleles for whatever reason, there should
be only a few common susceptibility alleles at each risk
locus, and the CDCV model should be true.
Neutral evolution predicts that alleles will be somewhat
less common than they would if governed by balancing
selection. If neutral susceptibility alleles happened to be
common in ancestral human populations, they should
still be common today (Reich & Landers 2001).
However, as noted earlier (sect. 4.1), genetic drift in
small ancestral human populations tends to drive neutral
loci to fixation, through random sampling error. Indeed,
most neutral loci seem to have one predominant allele
and, due to the recent increase in human population
size, many individually rare alleles, although some
neutral loci also have common alleles (Cargill et al. 1999;
Halushka et al. 1999). Thus, neutral evolution should
lead to a situation somewhere between the CDCV
model and the CDRV model.
Widespread mutation-selection, on the other hand,
should lead to a world where the CDRV hypothesis is
true. A trait’s genetic variation should be a result of
mutations at many different loci. The more deleterious
and common the trait was ancestrally, the more loci
would have to be involved; very serious and common
mental disorders may be affected by hundreds or even
thousands of potential loci, but only a portion of these
should contribute to the bulk of standing genetic variation
in any given population at any given time (Pritchard 2001).
At each locus, numerous different mutations should exist,
Keller & Miller: Resolving the paradox of heritable mental disorders
none of which should be at high frequencies (e.g., minor
allele frequencies of less than 5%). However, in cases
where selection against susceptibility alleles has been
minute (e.g., s , 1/5,000), such as might occur in the
case of gene-by-environment interactions (sect. 4.4), some
susceptibility alleles could be at high frequencies, despite
selection, due to random genetic drift (Pritchard 2001).
The historical success or failure of psychiatric gene
hunting helps clarify which of the three evolutionary
models – ancestral neutrality, balancing selection,
mutation-selection balance – best explains the existence
of the bulk of susceptibility alleles. The CDRV model,
most consistent with mutation-selection, predicts the
least progress in psychiatric gene hunting; whereas the
CDCV model, most consistent with balancing selection,
predicts the most. Where does the evidence stand? Once
again, mutation-selection seems to best fit the evidence.
Only a handful of replicable susceptibility alleles for
mental disorders have been found despite two decades
of intensive research involving thousands of scientists
and hundreds of millions of dollars. Acclaimed discoveries
of mental disorder susceptibility alleles have typically been
followed by repeated failures to replicate (Terwilliger &
Weiss 1998; Weiss & Clark 2002). At the same time, the
molecular bases for over 1,700 Mendelian phenotypes
have been definitively found to date (Online Mendelian
Inheritance of Man, April 10, 2006), showing that
current methods are wildly successful at finding alleles
responsible for single-gene, Mendelian disorders.
Even for these susceptibility alleles that have been
located, the effect sizes have been very small. One of the
more comprehensive recent meta-analyses (Lohmueller
et al. 2003) showed that only two of the eight moststudied mental disorder susceptibility alleles (at the
DRD3 and HTR2A loci) were reliably associated with a
mental disorder (schizophrenia). The meta-analysis
estimated the true odds ratio for the larger of the two
associations was just 1.12, meaning that given 1,000
people with the DRD3 susceptibility allele and 1,000
people without it, 11 people in the first group and 10 in
the second group will probably develop schizophrenia
(given its 1% base rate). Several other meta-analyses
have also recently concluded that discovered mental
disorder susceptibility alleles tend to have small effects
(odds ratios less than 1.5; Kendler 2005). The susceptibility alleles underlying most of the genetic risk for
mental disorders have not yet been found. If those that
have been found represent the “low-hanging fruit”
(explaining the most variation in the population), then
the remaining susceptibility alleles may be even rarer
and harder to detect.
We are not casting doubt on the entire enterprise of
gene hunting. Susceptibility alleles explaining the most
risk variation in the population, many of which may have
been found already, could be common because of balancing selection on separate traits, recent bottlenecks
among certain groups, or genetic drift (caused by fitness
effects that were closer to neutral ancestrally). If such susceptibility alleles happened to reach frequencies above 5%
in ancestral times, their current allelic complexity should
still be low, and gene-hunting techniques should be sufficient for finding them (Reich & Lander 2001). Some
protective alleles may be sweeping toward fixation caused
by recent selection. Some lineage-specific susceptibility
alleles may be missed within an analysis or not replicated
across analyses because of hidden population substructures that arose across evolutionary history. Technological
advancements may eventually enable discovery of even the
rarest susceptibility alleles, the base-pair sequences of
which would provide important information about the
relative importance of ancestral neutrality, balancing
selection, and mutation-selection balance (Bamshad &
Wooding 2003; Otto 2000). Nevertheless, the slow progress in finding mental disorder susceptibility alleles so
far, and the small amount of explained population risk of
those that have been found, are generally consistent with
the mutation-selection model and the CDRV model. If
balancing selection, and to a lesser degree ancestral neutrality, were general explanations for mental disorders,
then psychiatric genetics probably would have already
found the susceptibility alleles responsible for most of
the genetic variation underlying them.
8. Conclusions: Toward a resolution of the
paradox of common, harmful, heritable mental
Evolutionary anthropologist Donald Symons observed
that “you cannot understand what a person is saying
unless you understand who they are arguing with”
(Cosmides & Tooby, n.d.). In this article, we are arguing
mostly against those evolutionary thinkers who assume
that adaptive forces are the only possible explanations
for common, heritable polymorphisms such as mental disorders, even when those traits look profoundly harmful to
survival and reproduction. We are also arguing against
those psychiatric geneticists who disregard evolutionary
theory when trying to understand mental disorders
or their susceptibility alleles. This article has tried to
show how evolutionary genetic theories are important to
both fields.
Evolutionary psychologists have struggled to explain
genetic variation in the context of species-typical adaptive
design – sometimes ignoring it, sometimes citing mismatches between ancestral and current environments,
and sometimes trying to find hidden adaptive benefits
maintained by balancing selection. These approaches all
draw upon the familiar adaptive toolbox, in which the optimizing power of natural selection is assumed. This is a
great toolbox to use when trying to reverse engineer
universal aspects of human nature such as vision, mate
choice, or normal reactions to depression-inducing
situations. Indeed, the search for possible adaptive
functions of mental disorder symptoms, especially when
the capacity to express these symptoms is universal and
they are environmentally triggered, is an important
counterbalance to the prevailing assumption that subjective distress equals biological disorder. However, a very
different set of tools is required to explain persistent
genetic variation, especially in traits related to fitness.
These tools must be drawn from contemporary evolutionary genetics.
Psychiatric genetics has, with some pride, traditionally
been an empirically driven field. This approach is commendable to a degree. However, as Einstein once
observed, “It is the theory that decides what we can
observe,” and evolutionary genetics provides a rigorous
Keller & Miller: Resolving the paradox of heritable mental disorders
mathematical framework that could better guide
psychiatric gene hunting (e.g., Pritchard 2001; Reich &
Lander 2001; Rudan et al. 2003a; Wright et al. 2003).
For example, mutation-selection models suggest that
susceptibility alleles with the largest effect sizes may also
be the rarest, the most recent, and the most population
specific – an insight with important implications for the
methods most likely to locate mental disorder susceptibility alleles (see Wright et al. 2003). Moreover,
mutation-selection explanations further justify the search
for less polygenic, and more genetically mappable,
endophenotypes (Cannon & Keller 2005).
The existence of common, heritable, harmful mental
disorders creates an apparent evolutionary paradox, but
we think it can be resolved by recognizing the enormous
mutational target size of human behaviors. According to
this model, behavioral traits are especially susceptible to
harmful mutations because they depend on the most
complex organ in the human body. The brain is affected
by over half of the hundreds of mutations that all
humans carry. Some of these mutations have large,
distinctive effects, and so are reliably recognized as
Mendelian disorders. Tellingly, some of these mutations
cause syndromes inherited in Mendelian fashion but
that are otherwise phenotypically identical to mental
disorders (MacIntyre et al. 2003). Mendelian disorders
are rare because selection keeps very harmful mutations
very rare.
Most other mutations, especially in regulatory regions
of the genome, have much milder effects and cause
mostly quantitative differences in traits. Individuals with
an especially high load in mutations that disrupt a
particular configuration of brain systems will tend to act
in aberrant, harmful ways that provoke social comment
and psychiatric categorization. Lacking a map of the
neurogenetic watershed, psychiatrists have struggled to
identify criteria that could enable these behavioral syndromes to be meaningfully categorized. Current criteria
reflect perceived similarity of symptoms and prognoses,
which is potentially influenced not only by actual
etiological similarity, but also by the cultural and inherent
person-perception biases of those perceiving the sufferer,
and the categorization demands of legal, medical, and
research systems. Common mental disorders are
common because they are defined that way.
It was natural that these mental disorder categories
became reified, and that scientists looked for single
genes underlying them, which was so successfully accomplished with Mendelian disorders. But common mental
disorders are probably fundamentally different than
Mendelian disorders – not, as has often been presumed,
in that the former were not selected against while the
latter were – but rather, in that common mental disorders
are influenced by a much larger number of environmental
and genetic factors, most of which have only minor
influences on overall population risk.
Everyone alive, according to this model, has minor brain
abnormalities that cause them to be a little bit mentally
retarded, a little bit emotionally unstable, and a little bit
schizophrenic. If so, this framework may help explain
much more than just mental disorders; it may help
explain genetic differences between people in personality,
health, athleticism, intelligence, attractiveness, and virtually any other trait related to Darwinian fitness. If
scientists so chose, they could define the low-fitness
extremes of any of these dimensions as “disorders.” The
susceptibility alleles contributing to such “disorders”
would be the same ones responsible for genetic variation
across the whole dimension in the general population.
All other things being equal, someone of below-average
athleticism harbors an above-average number of
athleticism-degrading mutations. Adaptive organic
complexity is exquisite as an abstraction, but riddled
with errors within any living, breathing individual. We
are all very imperfect versions of that Platonic ideal, the
species-typical genome. This perspective may help
explain evidence of ubiquitous maladaptation; for
example, why nearly everyone suffers from some type of
heritable physical ailment, or why about half of people
will meet DSM-IV (Diagnostic and Statistical Manual
of Mental Disorders, 4th edition) criteria for a mental
disorder at some point in their lives (Kessler et al. 2005).
The theoretical and empirical evidence reviewed in
this article is most consistent with a polygenic mutationselection balance model for explaining common,
harmful, heritable mental disorders. Ancestral neutrality
and balancing selection almost certainly play roles in
maintaining some susceptibility alleles, but, as general
explanations, they are difficult to reconcile with empirical
evidence that mental disorders are associated with (1)
reduced fitness, (2) brain trauma, (3) higher paternal
age, (4) inbreeding, (5) genetic comorbidity, and (6)
many susceptibility alleles that explain little population
risk. So far, the evidence suggests that mutation-selection
plays an important role in maintaining susceptibility alleles
of mental disorders, whereas the other forces play less
certain roles. At the very least, we hope to have demonstrated that there is no necessary paradox in the existence
of common, heritable, harmful traits, such as mental disorders, and we hope to have shown the types of empirical
evidence that can test different evolutionary theories of
susceptibility alleles. It is possible, of course, that new
empirical evidence, or new understandings of how genes
affect phenotypes, will show that our conclusions were
substantively wrong. It is also possible that we have
made mistakes in interpretations of data or theory. This
is, after all, a persistent danger in multidisciplinary work,
but we feel strongly that the difficulties of integrating
such disparate fields are far outweighed by the potential
advantages. We look forward to a future in which
Darwinian psychiatry, psychiatric genetics, and evolutionary genetics become more mutually informative and
For helpful guidance and generous feedback, thanks to Paul
Andrews, Rosalind Arden, Nick Barton, Reinhard Bürger,
Tyrone Cannon, Greg Carey, Dan Fessler, A. J. Figueredo,
Steven Gangestad, Martie Haselton, Kenneth Kendler, Nick
Martin, Emily Messersmith, Michael Neale, Randolph Nesse,
Andrew Shaner, Srijan Sri, Kim Tremblay, Jerome Wakefield,
X. T. Wang, and Kenneth Weiss. The first author was
supported by a National Science Foundation Graduate
Research Fellowship, a fellowship from the UCLA Center for
Society and Genetics, and a National Research Service Award
from the National Institutes of Health, T32 MH-20030 (PI
M. C. Neale).
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
Open Peer Commentary
Praise for a critical perspective
David C. Airey and Richard C. Shelton
Department of Pharmacology, School of Medicine, Vanderbilt University,
Nashville, TN 37232-8548.
[email protected]
[email protected]
Abstract: The target article skillfully evaluates data on mental disorders
in relation to predictions from evolutionary genetic theories of neutral
evolution, balancing selection, and polygenic mutation-selection
balance, resulting in a negative outlook for the likelihood of success
finding genes for mental disorders. Nevertheless, new conceptualizations,
methods, and continued interactions across disciplines provide hope.
The insightful and balanced treatment in the target article takes a
pessimistic view of the likelihood of finding genes for mental disorders. In the view of the authors, if polygenic mutation-selection
balance truthfully characterizes the complex genetics of mental
disorders, our task may be more difficult than we had hoped,
barring new approaches. Nevertheless, several points in the
article resonated with us, including (1) an appreciation for evolutionary genetics, (2) the watershed analogy, (3) the concepts of
endophenotype and of genetic correlation, (4) the implied necessity of animal models, and (5) the need for cross-fertilization
among disciplines. We touch on each of these aspects in our
The authors’ analogy of the neurogenetic watershed is simplistic but useful. One of us (Shelton) has created and used a genetic
reference population of cultured human fibroblasts to investigate
intracellular signal transduction cascade differences in patients
with mood disorders. Using this system, it was discovered that
patients with the melancholic subtype of major depression have
reductions in the activity of protein kinases A and C (critical to
the synthesis of, for example, brain-derived neurotrophic factor
and the glucocorticoid receptor), and altered serotonin receptor
2A – mediated phosphoinositide signaling (Akin et al. 2004;
2005). These findings have been demonstrated in human postmortem brain tissue, as well. This research illustrates the ex
vivo study of endophenotypes very proximal to genes implicated
in depression (kinases, HTR2A), that is, very upstream in the
watershed model. Another approach, still in humans, uses
“imaging genetics” to study endophenotypes more downstream
at the level of neural systems. For example, genetic variation in
important serotonin genes (TPH2, the gene for tryptophan
hydroxylase, and SLC6A4, the serotonin transporter) has been
associated with activation of the amygdala by fearful stimuli
(Brown et al. 2005; Pezawas et al. 2005). Notably, in both
approaches, relatively smaller samples were required to detect
differences, implying genetic simplification followed the phenotypic simplification.
While the authors point out that heritability refers only to the
role of genetic variation in differences between individuals, and
not the similarities, they also highlight the usefulness of the
genetic correlation. A genetic correlation between two measured
traits implies shared allelic variants or genes in linkage disequilibrium. From an evolutionary standpoint, it is usually the former
that is more interesting (Airey et al. 2000) as common developmental programs are implicated. The genetic correlation is a
heavily used concept and tool in mouse genetics (Crabbe
1999). We look forward to the further development of this
concept in psychiatric genetics coupled with measured
The explicit emphasis on endophenotypes as measured
(continuous) traits, and the arguments against expectations of
distributional bimodality in affected (diagnosed) and unaffected
(undiagnosed) individuals, were welcome. Although perhaps
not the usual medical model, continuous variation opens the
door to a more complete understanding of mechanisms underlying individual differences, broadly defined. To be fair, psychiatric genetics has made significant inroads, showing that genetic
variation in psychiatric candidate genes like DAT, COMT,
TPH2, and SLC6A4 is associated with cognitive and emotional
dimensions in the normal human range (Bertolino et al. 2006;
Brown et al. 2005; Mattay et al. 2003; Pezawas et al. 2005).
We would like to point out that researchers focusing on other
complex genetic illnesses have successfully used populationbased sampling schemes to support the common disease rare
variant hypothesis. Cohen et al. (2005) found enrichment of
rare nonsynonymous variants of large effect size in candidate
genes for plasma levels of high-density-lipoprotein (HDL)
cholesterol from the lower 5% tail of a large population-based
sample. More recently, and in line with the target article’s predictions, Cohen et al. (2006) have found enrichment of rare variants
of a gene involved in cholesterol absorption, but where the effect
sizes of the variants were more moderate. Certainly, this leaves
open the possibility of effect sizes that are smaller still, as the
target article predicts. These works represent reasonable
models for the study of the genetics of mental disorders. A downside to this approach is that it is candidate gene based. Because
the approach relies on sequencing and polymorphism discovery,
rather than on typing known markers, the approach is not feasible
for genome scans. In other words, the onus is on the investigator
to understand enough about the biology of their disease to
nominate candidate genes for rare variant discovery.
The study of endophenotypes is presented as integral to progress despite the implications of polygenic mutation-selection
balance. However, it cannot be avoided that the study of endophenotypes in humans is limiting. Most neural endophenotypes
remain inaccessible. Therefore, genetic model organisms can
be used to widen the scope of understanding universal (translatable) brain mechanisms and to nominate candidate genes for
the aforementioned nonsynonymous rare variant methods.
The target article skillfully evaluates data on mental disorders
in relation to predictions from evolutionary genetic theories of
neutral evolution, balancing selection, and polygenic mutationselection balance. Although the arguments for the latter theory
are persuasive, we think it best, as Keller & Miller (K&M)
have done, to continually and critically evaluate theories that
are borrowed into secondary fields.
Genes for susceptibility to mental disorder are
not mental disorder: Clarifying the target of
evolutionary analysis and the role of the
Nicholas B. Allena and Paul B. T. Badcockb
Department of Psychology and ORYGEN Research Centre, University of
Melbourne, Parkville, 3010 Australia; bDepartment of Psychology, University of
Melbourne, Parkville, 3010 Australia.
[email protected]
[email protected]
Abstract: In this commentary, we critique the appropriate behavioural
features for evolutionary genetic analysis, the role of the environment,
and the viability of a general evolutionary genetic model for all
common mental disorders. In light of these issues, we suggest that the
authors may have prematurely discounted the role of some of the
mechanisms they review, particularly balancing selection.
Keller & Miller’s (K&M’s) integration of quantitative evolutionary genetics with evolutionary approaches to psychiatric disorder
is timely and important. Their article convincingly argues that the
role of polygenetic mutation-selection balance has not been
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
adequately recognised within evolutionary psychiatry. Although
we are generally in agreement with their argument, we would
like to offer critical commentary on three issues.
The first concerns determining appropriate behavioural
features for evolutionary analysis. Much of K&M’s argument is
framed in terms of the concept of mental disorder, although
they also provide an incisive critique of the concept. However,
their treatment is often inconsistent in terms of the phenomena
it considers. The evidence they review is usually relevant to
case-level mental disorder, but, in places, they focus on variations
in underlying susceptibility traits, and, in others, on genes that
confer susceptibility for mental disorder. Our own view is that
evolutionary genetic analysis should not focus upon mental disorders per se, but on genetic factors that underlie susceptibility
to disorders – on genes that result in extreme variation in
polygenetic traits, and on how such variation causes individual
differences in the functioning (e.g., threshold of activation, propensity for regulation/dysregulation) of evolved psychobiological
mechanisms. Heritable variation in such traits can be maintained
by a range of factors that potentially include each of those
discussed by K&M, depending on the trait.
This point of view results in the following agenda for genetic
evolutionary psychiatry. First step is an analysis of the adaptive
basis of a species-typical psychobiological mechanism, including
the recurring ancestral adaptive problem(s) that it solved, and
how the design of the feature provided an efficient, domainspecific solution to the problem(s) (Buss 1995). Such analyses
are typical in evolutionary psychology, an example being our
own work on depressed mood (see Allen & Badcock 2003).
Second, there should be an analysis of the likely sources of
genetically maintained variation in the functioning of the trait.
Finally, the theorist should consider the circumstances under
which the operation of the mechanism is dysfunctional (i.e., no
longer performing its naturally selected function; Wakefield
Such an agenda has implications for the issues considered by
K&M. For example, a central concern is why genes that confer
susceptibility to mental disorder are maintained by natural selection when such disorders are patently maladaptive. As evidence
for this claim, K&M refer to findings that fertility rates
amongst those suffering from case-level disorders are persistently
lower than for non-psychiatric populations. The approach just
outlined, however, would suggest that their analysis should
instead concentrate on the socio-reproductive consequences of
trait variation in the functioning of evolved mechanisms. For
example, personality is a salient descriptor of individual
differences in susceptibility to mental disorder (Clark 2005;
Kruger & Tackett 2003), and can be thought of as representing
phenotypic differences in the functioning of evolved mechanisms
for dealing with, for example, social and environmental threat
(neuroticism; Nettle 2004) or seeking out propitious environments (extraversion; Depue & Collins 1999). A study of personality and reproductive fitness in humans found that optimal
levels of fertility were observed at two loci: those displaying
high neuroticism and low extraversion (a group who are likely
to have susceptibility genes for mental disorder), and those
displaying high extraversion and low neuroticism (Eaves et al.
1990). Such data indicate that although mental disorders
themselves may be associated with low fertility, the genes that
confer susceptibility to these disorders may not. K&M’s
dismissal of the ancestral neutrality and balancing-selection
models may therefore have been premature for some disorder
Our second contention is that K&M appear to underestimate
the role of environmental influences. As noted by others (e.g.,
Hall 1999; Lickliter & Honeycutt 2003; Oyama et al. 2001),
the developmental outcome of the self-regulating, multileveled
system that characterises an organism (and the more specific
psychobiological design features it exhibits) is not prescribed
by the genes alone, but by the regulatory dynamics of a
complex, gene-in-a-cell-in-an-organism-in-an-environment system.
For example, recent research has demonstrated that selection
for a personality trait with high heritability fluctuates across
years within a natural bird population, suggesting that
fluctuations in gene-environment interactions can maintain
genetically specified personality variation (Dingemanse et al.
2004). These data imply that even relatively minor variations
within environments may result in the maintenance of an allele
in the population despite its deleterious effects in particular
contexts. Indeed, for most of the common mental disorders
analysed by K&M, strong gene-environment interactions in
their aetiology are the rule rather than the exception (e.g.,
Caspi et al. 2003; Kendler & Eaves 1986). Thus, when phylogenetic explanations focus on susceptibility alleles that underlie
endophenotypic trait deviation in the functioning of adaptive
mechanisms – rather than disorders per se – ancestral neutrality
and balancing-selection models become more plausible.
Third, and finally, we believe that searching for a general set of
evolutionary genetic models that explain all common mental
disorders, irrespective of their specific features, may be a
fraught goal. Disorders differ in prevalence, heritability,
whether they are continuous or discontinuous with normal functioning, and the importance of environmental precipitants –
both in terms of their variation in modern ecologies, and the
influence of historical changes in environments (such as those
between ancestral and recent times). These factors, amongst
others, will help adjudicate the likely role of ancestral neutrality,
balancing selection, and polygenetic mutation-selection balance
in the maintenance of susceptibility alleles. As such, it may be
more fruitful to examine these issues within each disorder (or,
more specifically, in terms of maladaptive deviations from
well-defined adaptive psychobiological mechanisms), than for a
general class of disorders. Indeed, committing to the power of
any one explanatory model to account for the full diversity of
psychopathologies (and the complex interplay of phylogenetic
and ontogenetic processes responsible for each) invariably runs
the risk of neglecting alternate, equally viable, and possibly
complimentary hypotheses. Nevertheless, we welcome K&M’s
contribution to the literature and trust that it will enkindle a
more rigorous and scientifically principled development of
models within evolutionary and genetic psychiatry.
1. By “no longer performing its naturally selected function,” we mean
that the action of the mechanism either exacerbates the problem(s) it was
designed to solve or creates a new problem that is more severe or
The social environment compresses the
diversity of genetic aberrations into the
uniformity of schizophrenia manifestations
Ralf-Peter Behrendt
MRC Psychiatry, The Retreat Hospital, York, YO10 5BN, United Kingdom.
[email protected]
Abstract: Genetically and neurodevelopmentally, there may be a
thousand schizophrenias, yet there would be no schizophrenia at all
without active contribution from all of us; none – outside the primitive
processes that regulate our relationship with one another. In order to
understand the nature of schizophrenia as it unfolds relatively
uniformly in the social context, we need to depart from an
evolutionarily more feasible understanding of society.
As long as we are sane and he is insane, it will remain so. But comprehension as an effort to reach and grasp him, while remaining within our
own world and judging him by our own categories whereby he inevitably falls short, is not what the schizophrenic either wants or requires.
We have to recognize all the time his distinctiveness and differentness,
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
his separateness and loneliness and despair. . . . Schizophrenia cannot
be understood without understanding despair.
—R.D. Laing (1960, p. 39)
In Keller & Miller’s (K&M’s) thorough exposition, one conjecture
remains unconvincing. The question as to how the multitude of
susceptibility alleles – maintained by mutation selection and
expressed in a multitude of neurodevelopmental “imperfections” – channels into discrete disorders such as schizophrenia
cannot simply be sidestepped by claiming that “mental disorder
categories became reified” (sect. 8, para. 6) and the “similarity
of symptoms” is merely “perceived” (sect. 8, para. 5). Let us try
to reconcile their elegant model with the fact that schizophrenia
does, after all, represent a fairly separate entity, a unitary disease,
that is not primarily the product of “historical convention,” “diagnostic convenience” (sect. 6.4, para. 6), or requirements “to make
certain social decisions” (sect. 6.4, para. 5). Let us ask how nonspecific deficits in adaptability can translate into the rather
specific phenomenology of psychotic breakdown experienced
episodically by schizophrenic patients (who do not, by the way,
“imagine hostile, confusing voices” [sect. 1, para. 1] but actually
hear them). I argue that K&M’s model, which emphasises the
etiological and neurodevelopmental heterogeneity of schizophrenia, does not necessarily imply that schizophrenia cannot
be a “natural kind” (sect. 6.4, para. 5). Although underneath
schizophrenia there may “lie a vast diversity of potential
behavior-impairing mutations across the thousands of genes
involved in brain development” (sect. 6.4, para. 6), the manifest
homogeneity of schizophrenia may still be real. The argument
is that “individuals with an especially high load in mutations”
who “act in aberrant, harmful ways” do indeed “provoke social
comment” (sect. 8, para. 5). But it is precisely the type of
response from the social environment to such aberration that
shapes an individual’s mental illness into what then tertiarily provokes “psychiatric categorization” (sect. 8, para. 5) as schizophrenia.
Intraspecific aggression provides the motivational impetus to
social structures and group dynamics in many higher vertebrates,
including humans, although our desire to regard our species as
standing above the animal kingdom prevents us from fully appreciating this (Lorenz 1963/2002). Through ritualisation in phylogenesis and cultural evolution, aggressive impulses arising
spontaneously in social situations became partly inhibited and
adopted indirect expressions, thus contributing to the wide
range of social behaviours (Lorenz 1963/2002; see also Behrendt
2006). We need to relate to others to keep at bay existential
anxieties (that is, anxieties originally experienced by the infant
who comes to realise his existential dependence on his caregiver),
yet in groups we are constantly faced with others’ concealed
readiness to attack. Unless we prefer the insecurity and existential anxiety that loneliness makes inescapable (or escapable only
in one’s unconscious “omnipotent” phantasy), we have to
develop complicated, adaptive behaviour patterns aimed at
appeasement, submission, and assertion that in fluctuating
group constellations respond to a multitude of cues indicative
of others’ social ranking, aggressive potential, and intentions. It
is important to understand social conformity essentially as a
mechanism that deflects aggressive impulses naturally arising
from the social environment with which we engage. If we fail
to negotiate our social rank successfully under changing circumstances, and if we fail to maintain the most complex etiquette and
attitudes which we have learnt are demanded by the situation, we
will face rejection and resurging existential anxieties (against
which we have to develop elaborate defensive systems that themselves may test to the limit higher cognitive abilities).
Indeed, many mechanisms can “disrupt adaptive complexity”
(sect. 7.2) and thus undermine our ability to conform. Schizophrenia is not just a “heritable” disease, as K&M recognise, but
more generally a neurodevelopmental disorder that apart from
mutation loading is associated with prenatal exposure to viral
infections (Brown & Susser 2002), obstetric complications
(Verdoux et al. 1997), and childhood brain disease, manifesting
not only in abnormal brain development with morphological
and cytoarchitectonic alterations, but also in postural and movement deficits in childhood (Walker et al. 1994), minor physical
abnormalities, cognitive impairment, and developmental delays
(Jones et al. 1994). Although it is often argued that aberrant
early development is part of the schizophrenic process, it is
clear that if children or adolescents thus affected lack compensatory strengths and alternative sources of support, they may be
placed on a trajectory of social rejection, fears of rejection,
social maladjustment, and unappeasable existential anxieties
that may lead to and – at times of additional social stresses and
insecurities – clinically present as schizoid or schizotypal,
perhaps paranoid, personality disorder or, indeed, schizophrenia – if the individual has a hallucinatory predisposition in
addition (Behrendt & Young 2004).
Compliance with social rules and others’ expectations is
designed to keep in check others’ aggression, yet conformity is
precisely what is difficult to achieve – for various reasons – by
those who are at risk of developing schizophrenia, allowing existential anxieties to prevail. Laing (1960) described the schizoid
individual as having a “heightened sense of being always seen,”
as being “frightened that he will look a fool, or that other
people will think he wants to show off” (pp. 113– 14):
In a world full of danger, to be a potentially seeable object is to be constantly exposed to danger. . . . Indeed, considered biologically, the very
fact of being visible exposes an animal to the risk of attack from its
enemies, and no animal is without enemies. Being visible is therefore
a basic biological risk; being invisible is a basic biological defence.
We all employ some form of camouflage. (Laing 1960, p. 117)
What better camouflage, what better way to avert attack, than
our “false self” or persona representing our strivings for conformity? Defences employed against the “many anxieties about
being obvious, being out of the ordinary, being distinctive,
drawing attention to oneself” – in essence representing fears of
rejection – “often consist in attempts to merge with the human
landscape, to make it as difficult as possible for anyone to see
in what way one differs from anyone else” (p. 118). For the schizoid individual to enter into essential exchange with others means
to lay himself open to attack, insofar as he cannot conform. He
maintains “his outward semblance of normality by progressively
more desperate means” until his “defences against the world
fail even in their primary functions: to prevent persecutory
impingements” so that anxiety “creeps back more intensely
than ever” (p. 150). Alternatively, “in order to be safe from the
persistent threat and danger from the world,” he has to cut
himself off “from direct relatedness with others” (pp. 149– 50).
Withdrawal into “omnipotent” phantasy affords transient lowering of existential anxieties; however, this comes at a price of
increasing estrangement from the shared reality, eventually
driving the individual into psychosis.
Children who later develop schizophrenia are more anxious at
school and more likely to play alone (Jones et al. 1994). Poor
social adjustment and lack of confidence are common characteristics in children and adolescents who later develop schizophrenia (Jones et al. 1995) but may be nothing more than
consequences of sustained subtle or overt aggression from
peers. Vulnerable children are at risk of attracting others’ aggression quite naturally through their being slow, weak, unattractive,
or simply different, and this would be even more so if they were
anything but content with the lowest rank in their peer group.
Failure by individuals to conform disinhibits our instinctive
aggression, which is why those who are about to develop schizophrenia start to observe their social environment fearfully,
especially at times when parents’ and society’s expectations as
to the individual’s ability to conform and fulfil acceptable roles
is greatest – that is, in puberty and early adulthood. The
association of schizophrenic relapse with exposure to criticism
and hostility in a patient’s family (Brown et al. 1972) and the
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
association with urban upbringing (Pedersen & Mortensen
2001) – which implies instability of links with social groups –
attest to the importance of intraspecific aggression and our fear
thereof in schizophrenia; as do hallucinations of other people’s
voices during psychosis, which in their content are often critical
and derogative (Birchwood et al. 2000; Linn 1977), reflecting
patients’ fears of rejection and marginalisation.
Evolutionary psychiatry is dead – Long liveth
evolutionary psychopathology
Martin Brüne
Centre for Psychiatry, Psychotherapy, and Psychosomatics, University of
Bochum, 44791 Bochum, Germany.
[email protected]
Abstract: Keller & Miller (K&M) propose that many psychiatric
disorders are best explained in terms of a genetic watershed model.
This view challenges traditional evolutionary accounts of psychiatric
disorders, many of which have tried to argue in support of a presumed
balanced polymorphism, implying some hidden adaptive advantage of
the alleles predisposing people to psychiatric disorders. Does this mean
that evolutionary ideas are no longer viable to explain psychiatric
disorders? The answer is no. However, K&M’s critical evaluation
supports the view that psychiatric disorders are not categorically
distinct from normalcy, and that evolutionary psychopathology should
be grounded in rigorous empirical testing.
Keller & Miller’s (K&M’s) proposal of how to best conceptualize
the genetic underpinnings of psychiatric disorders will certainly
induce an uproar among those evolutionary psychiatrists who
believe that psychiatric disorders reflect adaptations to selectively
relevant problems in the human evolutionary history. Adaptive
properties have been assigned to a great diversity of psychiatric
problems, such as anxiety disorders, depression, and even
vaguely defined disorders such as schizophrenia, either on
account of their mere prevalence in the population or due to a
presumed function of susceptibility alleles in heterogeneous
carriers (for a critique, cf. Dubrovsky 2002). In contrast, K&M’s
suggestion that many psychiatric disorders emerge if the number
of “unfavourable” alleles is sufficiently great in an individual,
thereby rendering this individual vulnerable to environmental
stress, challenges the adaptationist perspective on psychiatric
disorders, because it would seem that the search for genes
predisposing people to psychiatric disorders and for a hidden
adaptive advantage of such genes could be a fruitless endeavour.
Does K&M’s account discredit an evolutionary approach to
psychopathology altogether? My answer is no. However,
K&M’s thoughtful review of possible genetic models of psychiatric disorders unveils several weaknesses of the adaptationist
perspective in psychiatry.
First, the adaptationist perspective holds that some psychiatric
disorders, while perhaps being maladaptive themselves, convey
adaptive advantages in first-degree relatives of the affected individuals outweighing the reproductive disadvantage of the
symptom carriers, similar to the oft-cited example of sickle-cell
anaemia. Such a hypothesis could, however, be empirically
tested only if such disorders were well defined at the phenomenological, neurophysiological, and genetic levels (sickle-cell
anaemia is monogenetically transmitted). This is not the case
for almost all psychiatric disorders, as there is a broad overlap
between psychiatric disorders at all levels. For example, abundant research now points to the fact that schizophrenia and
bipolar affective disorder share several clinical features (e.g.,
thought disorders) and neurophysiological characteristics
(e.g., elevated dopamine levels), and run in the same families
(i.e., genetic susceptibility).
Second, some evolutionary psychiatrists assert that psychiatric
disorders themselves were adaptive responses to adverse
environmental conditions. For instance, depression could
reduce social stress by causing a person to assume a subordinate
behavioural strategy instead of fighting for resources, if the
chance of succeeding in contest is perceived to be low. This
view is much more delicate, because it is plausible for minor
forms of reactive depression, for example, after loss of social
status or after divorce. It becomes, however, much more difficult
to explain severe, melancholic, or “endogenous” depression,
which, rather than being adaptive, appears to be similar to
what ethologists call vacuum behaviour (behaviour occurring
without appropriate stimulus). This example may illustrate the
difficulty in drawing a line between an adaptive “second-best”
strategy and maladaptive behaviour, based on severity of the
symptomatology or the presence or absence of stimuli (Nettle
2004). In any event, it needs to be emphasised that sadness is
not depression, and that suspiciousness does not equal delusional
thinking, although in both examples the former may develop into
the latter. The problem of continuity between normalcy and
pathology has not yet been acknowledged by many evolutionary
psychiatrists, perhaps because the majority of them still think
in categorical dimensions.
Ironically, K&M’s review of the possible genetic underpinnings of psychiatric disorders from an evolutionary perspective
strengthens the view that environmental factors are perhaps at
least equally important for explaining psychopathology in evolutionary terms (this is not to say that genetic variation does not
play a role in psychopathology). If it is true that every one of us
carries several hundreds of “unfavourable” alleles rendering all
of us more or less susceptible to developing psychiatric disorders,
it would be much more advisable to carefully investigate potentially unfavourable environmental conditions eliciting psychopathology. For example, Belsky et al. (1990) have proposed an
intriguing model of how behaviour rendered “deviant” in
modern environments may emerge if infants grow up under
adverse conditions, including harsh parental rearing styles, lack
of sufficient resources, and so on. The model basically suggests
that, under favourable circumstances, parents could afford to
invest heavily in their offspring, hence promoting in children
the development of an inner working model of trustworthiness;
whereas it might also have been adaptive for offspring in
adverse ancestral conditions to assume a view that the world is
an insecure and unpredictable place, where (in terms of reproductive fitness) it was advisable to focus on immediate resource
extraction, including the exploitation of others. From a psychopathological perspective, the extremes of the latter “strategy”
(no conscious reflection implied) may fall into the categories of
antisocial personality disorder (ASPD) or borderline personality
disorder (BPD), both of which can be seen as pathological (and
therefore maladaptive) exaggerations of an adaptive mechanism.
Another strength of evolutionary psychopathology is certainly
to help clarify sex differences in behaviour. With respect to the
foregoing example, ASPD is much more common in males,
whereas BPD is more often diagnosed in females, although the
precipitating condition may be analogous or even identical.
This might reflect sex-dependent competitiveness in different
social arenas. Similarly, traditional psychiatry has been unable
to explain sex differences in the content of delusional disorders,
where evolutionary psychopathology can offer testable hypotheses about why erotomania reflects the pathological exaggeration
of a female mating strategy, whereas delusional jealousy reflects
the pathology of a male mating strategy (Brüne 2003). There is no
psychiatric framework other than evolutionary psychopathology
to explain such differences between the sexes (Brüne 2002).
In summary, evolutionary psychopathology (as a field) would
do better if it chose a more symptom-based or syndromal
approach rather than clinging to nosological categories that
have proven more and more outdated – except for the purpose
of bringing evolutionary issues of psychopathology to a broader
audience of psychiatrists and clinical psychologists. This is one
important message from K&M’s thought-provoking thesis.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
The evolutionary genetics of personality: Does
mutation load signal relationship load?
David M. Buss
Department of Psychology, University of Texas, Austin, TX 78712.
[email protected]
Abstract: The mutation-selection hypothesis may extend to
understanding normal personality variation. Traits such as emotional
stability, agreeableness, and conscientiousness figure strongly in mate
selection and show evidence of non-additive genetic variance. They are
linked with reproductively relevant outcomes, including longevity,
resource acquisition, and mating success. Evolved difference-detection
adaptations may function to spurn individuals whose high mutation
load signals a burdensome relationship load.
Keller & Miller (K&M) must be applauded for a brilliant article
that provides the most compelling theory we now have for the
evolutionary genetics of many mental disorders. This commentary extends the logic of their arguments to the evolutionary
genetics of normal personality dimensions.
Mental disorders originating from a high mutation load,
according to K&M, undermine reproductive success primarily
by reducing the mating attractiveness of those afflicted. A
similar argument can be made for normal personality variation.
The ends of some personality dimensions are known to be
highly desirable in long-term mates (Buss & Barnes 1986;
Buss et al. 1990). A study of 37 cultures found that, after
“mutual attraction and love,” “dependable character” (conscientious) and “emotional stability and maturity” were the most
highly valued among 18 characteristics rated for their desirability
in a long-term mate. Using a ranking procedure, “kind and
understanding” (synonyms for agreeableness) and “intelligent”
topped a list of 13 characteristics as the most desirable in a
mate. The high ends of many major dimensions of personality,
in short, figure importantly in an individual’s “mate value”
(Buss 2003).
Recent evidence reveals that major personality traits are
polygenic and show substantial non-additive genetic variation
(Eaves et al. 1998; Keller et al. 2005). These findings are consistent with the polygenic mutation-selection hypothesis.
Whereas mental disorders impair reproductive success, the
positive ends of attractive personality traits facilitate fitnessrelevant outcomes (Buss & Greiling 1999). Those high in conscientiousness, for example, tend to excel in resource acquisition
and ascend status hierarchies (Kyl-Heku & Buss 1996; Lund
et al., in press). They also live longer (Friedman et al. 1995).
Those low in emotional stability (high on neuroticism) have
greater difficulty holding jobs and sustaining marriages.
Whether these difficulties historically undermined, or currently
undermine, reproductive success remains unknown.
If a person’s overall mate value is comprised of many different
personality traits (along with non-personality variables such as
physical attractiveness), then a given level of mate value can be
attained through different combinations of personality traits
(Buss & Barnes 1986). If people mate assortatively based on
overall mate value, rather than on individual personality traits,
one would expect low but positive assortment coefficients for
the individual personality traits. That is precisely what assortative
mating studies reveal (Buss 1984). Another prediction that
follows is that higher assortative mating coefficients should be
obtained by creating composite measures, summing the scores
of each of the different desirable personality traits. High
scorers – individuals with a highly attractive composite mating
personality – should attract other high scorers. Those of lower
mate value settle for commensurately lower-value partners.
This prediction remains to be tested.
All of these findings – the importance of personality in mate
selection, the importance of non-additive genetic variance for
personality traits, the links between personality and reproductively relevant outcomes, and the low positive assortment
coefficients for personality – are consistent with, but do not
definitively verify, the hypothesis that mutations degrade personality performance. The arguments of K&M in the context of
mental disorders, however, logically extend to the realm of
normal personality functioning, at least for personality
dispositions central to solving critical adaptive problems.
Evolved personality-assessment mechanisms (Buss 1996) –
the categories we use to appraise and evaluate others in our
social world – may function, in part, to assess the mutation
load carried by potential mates. They might also function to
assess the quality of allies, coalition partners, or even children
who might be especially attractive targets of parental investment.
These evolved difference-detecting adaptations provide
answers to some of the most important social problems that
people have faced while traversing the adaptive landscape
(Buss 1996): Will X be a good cooperator (agreeableness)? Will
X be a hard-working resource provider and reliable in provisioning my children (conscientiousness)? Does X have the fortitude
and resilience to hold steady during times of trouble (emotional
stability)? Conversely, will X cause psychological damage (aggressiveness), betray my trust (impulsiveness), or neglect my children
Individuals with certain personality characteristics inflict
fitness costs on those with whom they become enmeshed.
People low on emotional stability and low on agreeableness, for
example, create tremendous conflict in mating relationships,
absorbing valuable fitness-relevant resources that could be
better allocated elsewhere (Buss 1991). In these ways, mutation
load may signal relationship load.
Finland’s Galapagos: Founder effect, drift, and
isolation in the inheritance of susceptibility
Tom Campbell,a Daria Osipova,b and Seppo Kähkönenb,c
Helsinki Collegium for Advanced Studies, University of Helsinki, FIN-00014
Helsinki, Finland; bCognitive Brain Research Unit, University of Helsinki,
FIN-00014 Helsinki, Finland; cBioMag Laboratory, Helsinki University Central
Hospital, FIN-00290 Helsinki, Finland.
[email protected]
[email protected]
[email protected]
Abstract: The target article excludes ancestral neutrality as a cause for
the inheritance of schizophrenia, with an argument relating to selection
against a single allele in the Finnish population. However, drift would
predominate over selection within subisolates of the Finnish
population. Comparisons of subisolates with heterogeneous populations
may provide clues to the endophenotypic structure of complex
polygenetic heritable mental disorders.
It is paradoxical that, though individuals who have mental disorders are more likely to die childless, the predisposition to
these disorders seem to be inherited and the incidence rates
seem not to decline across generations. Natural selection still
has not uniformly eliminated the sets of gene codings in human
DNA, the susceptibility alleles, which predispose some humans
to mental disorder. The target article contrasts three extant
models of the evolution of mental disorder. These models offer
solutions to this apparent paradox, and this commentary discusses these models in the context of schizophrenia and autism.
The first of these models, ancestral neutrality, assumes that,
within the environment of our evolutionary ancestors, the selection disadvantage of predisposition to mental disorder present in
modern environments was absent. Accordingly, it is just a matter
of time before the susceptibility alleles become eliminated.
The second balancing-selection model assumes that, although
mental disorder has a selection disadvantage, that disorder also
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
offers selection advantages. An instance of this model is that
schizophrenia is the price that humans pay for the single allele
that determines the lateralization of function that permits
language (T. J. Crow 2000).
The third, favoured, polygenetic mutation-selection balance
model offers a view that predisposition to mental disorder has
no selection advantage. Rather, inheritance of mental disorder
reflects harmful germ-line mutations. Large numbers of these
mutations are a necessary by-product of the development of a
complex organism, and the deleterious effects of these mutations
derive from multiple mutant alleles at different genetic loci. This
ties in well with the assumption that neural systems supporting
mental processes and behaviour are thus under polygenetic
control. Accordingly, the human brain can be affected by more
than half of the hundreds of mutations that we all carry.
Although a single allele can result in a Mendelian disorder with
near disastrous consequences such that selection pressures
nearly eliminate their incidence (see Table 1), the mutation of
just a single gene tends to be relatively benign. These mutations,
when they bring modestly deleterious effects, do not severely
impair reproductive success and therefore are not strongly
selected against. The consequence is that such mutant alleles
can produce a sub-threshold abnormality that can be inherited.
The target article excludes the alternative ancestral neutrality
model in the context of schizophrenia (sect. 4.3). Keller &
Miller’s (K&M’s) argument here considers a hypothetical schizophrenia in Finland, determined by one or even a few susceptibility alleles, to demonstrate that if selection pressure against
this gene were the same as it is now, 42% of Finns would have
been schizophrenic in 1600. Whilst this argument remains compelling, it would be interesting to see how well it holds by contrast
to a population other than the Finnish one.
The Finnish population would seem a curious choice to use in
this argument, because in some regards the population is unique.
Factors such as the small number of initial founders, famines and
war, and rapid expansion during the last 80 – 100 generations
have left Finland with an anomalous genetic background
(Auranen 2000). Only the coastline of Finland and the southeast
and southwest areas were inhabited up until the 16th century.
The immigration of farmers to the inland established the late
settlement region in central Finland, creating small regional subisolates which have remained rather secluded from any further
immigration. Investigation of the genetics of these subisolates
has proved effective in identifying novel susceptibility alleles
for a number of conditions, including autism. This finding
supports the hypothesis of enrichment of different autismpredisposing alleles in the Finnish population, due to founder
effect, genetic drift, and isolation.
The heritability of autism is amongst the highest of the
common mental disorders (Table 1), whereas its prevalence is
amongst the lowest – which may lead one to think that autism
somewhat resembles the Mendelian disorders. Indeed, fewer significant risk loci have been enriched in the Finnish population
than in the larger and genetically more heterogeneous population
of the United States. However, indications in both populations
are that of a polygenetic disorder (Ylisaukko-Oja et al. 2006).
The first settlers to Kuusamo in north-eastern Finland arrived
from the late settlement region in 1676. Isolation from subsequent outbreaks of disease across more densely populated
areas permitted the faster expansion of this Kuusamo subisolate
relative to the rest of Finland: From just 40 founding families,
descended 18,000 people with triple the risk of a schizophrenia
of the national population, yet with comparable clinical phenotype (Arajärvi et al. 2004; Hovatta et al. 1999). Even this
Kuusamo subisolate exhibits multiple susceptibility alleles for
schizophrenia, corroborating the target article’s argument for a
polygenetic disorder (Hovatta et al. 1999).
However, the argument against an ancestral neutrality account
of schizophrenia excludes the effects of genetic drift (sect. 4.3;
see also the caption of Fig. 2). Within subisolate populations,
drift predominates over selection, particularly when the effects
of an allele are relatively benign. This argument against ancestral
neutrality would be more compelling were the authors to contrast
the assumed effects of drift and selection on this hypothetical
allele in heterogeneous populations and those containing
Ultimately, the third, favoured, polygenetic mutation-selection
balance account may offer a more convincing explanation of data
concerning heritable mental disorders. An individual allele may
manifest as a sub-threshold abnormality, an endophenotype,
such as impaired immediate memory or decreased efficiency of
dorsolateral prefrontal cortex (Bertolino et al. 2004; Campbell
2005; Egan et al. 2001; Tuulio-Henriksson et al. 2003), which,
though harmful, is neither necessary nor sufficient for disorder.
Expressed together, clusters of such endophenotypes,
however, may cause complex conditions like schizophrenia.
Particular endophenotypes might also be clinically useful in
determining the choice of drug (Bertolino et al. 2004). This
assumption of endophenotypes within the polygenetic
mutation-selection balance model thus offers something
additional that Crow’s balancing-selection account of schizophrenia would seem not to. Indeed, the correspondence of
endophenotypes of a disorder to susceptibility alleles in subisolate populations could also offer clues to the endophenotypic
structure of the disorder in more heterogeneous populations.
The natural selection of psychosis
Bernard Crespi
Department of Biosciences, Simon Fraser University, Burnaby, British
Columbia, V5A 1S6, Canada.
[email protected]
Abstract: Diverse evidence from genomics, epidemiology, neurophysiology, psychology, and evolutionary biology converges on simple
general mechanisms, based on negative secondary effects of strong
selection, for how mental disorders such as psychosis have evolved and
how they are sustained.
The core of my argument is that selection can explain the
observed genetic and phenotypic data on common mental
disorders. I focus on psychosis in general and schizophrenia in
particular. My thesis is predicated on a history along the
human lineage of selection for social cognition, creativity, and
language. I also contend that schizophrenia is a maladaptive
by-product of this strong selection. This general argument has
been made by Horrobin (1998) in the context of brain phospholipid metabolism, and by Crow (1997), Burns (2004), and Nesse
(2004) in the context of language and social intelligence. It translates into some combination of two selective processes, each with
conjoined positive and negative effects on phenotypes: (1) a
history of strong selection coupled with antagonistic pleiotropy
or linkage, and (2) the maintenance of variation via overdominance (balancing selection) across multiple loci. Three primary
lines of evidence support this argument.
First, many genes associated with psychosis have been subject
to positive selection – that is, selection for specific, favored
amino acid changes. Genes associated with schizophrenia or
bipolar disorder that show evidence for positive selection in the
human lineage include APOL1, APOE, DRD4, FOXP2,
SYNJ1 (Abdolmaleky et al. 2005; Andres et al. 2004; Balciuniene
et al. 2001; Costas et al. 2005; Diller et al. 2002; Ding et al. 2002;
Enard et al. 2002; Gardner et al. 2006; Harrison & Weinberger
2005; Jansson et al. 2005; Kitano et al. 2004; Muglia et al.
2002; Saito et al. 2001; Sanjuan et al. 2006; Spinks et al. 2004;
Stopkova et al. 2004; Voight et al. 2006; Yoshikawa et al. 2001;
Zhang et al. 2002). Overwhelming evidence from domesticated
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
species and laboratory experiments indicates that strong selection leads to maladaptive, more or less transient by-products,
on account of linkage and antagonistic pleiotropy in populations
out of equilibrium (e.g., see Andolfatto 2001; Lu et al. 2006).
These data suggest that selection for the traits that have “made
us human” (cf. Horrobin 1998), especially the neural systems
underlying language and social cognition, have led to psychosis
as a secondary result. Data are now available to test this hypothesis more directly, using the human haplotype map to test for
selective sweeps in regions associated in genome scans with psychosis, such as 1q21 (Voight et al. 2006). Many of the selective
sweeps inferred from such data (Voight et al. 2006) are remarkably recent (less than 20,000 years old). As a result, allele frequencies may be out of equilibrium, and equilibrium-based
population-genetic models for explaining standing levels of variation, based on antagonistic pleiotropy or related mechanisms, do
not apply.
Second, evidence for multilocus overdominance comes from
multiple studies showing increased fitness, compared to the
general population, in first-order relatives of schizophrenics.
The study by Haukka et al. (2003) found such an effect for
females, but not for males, and they cite four previous studies
supporting such a difference. A stronger pattern in females fits
with the less-debilitating nature of psychosis in this sex (Moriarty
et al. 2001), and such a sex bias was also found by Fananas and
Bertranpetit (1995) and Bassett et al. (1996). Nettle and Clegg
(2006) also report an association between increased mating
success and measures of schizotypy. The sample in Haukka
et al. (2003) is indeed very large, but no single such study can
be definitive or serve to estimate selective parameters quantitatively, given population-specific effects and the evolutionary time
scale involved. The upshot is that six independent studies support
a general pattern of balancing selection, apparently related to
positive aspects of schizotypy.
Third, there is substantial evidence for mechanisms that can
generate multilocus balancing selection on relevant aspects of
cognition. The causal links between measures of schizotypy and
measures of creativity and divergent thinking are much stronger
than Keller & Miller (K&M) imply, and they comprise diverse
evidence from functional imaging, neurophysiology, neural
network modelling, genomics, and psychological experiments,
as well as the biographical and survey studies discussed by
Waddell (1998) (Abraham et al. 2005; Brugger 2001; Fisher
et al. 2004; Folley et al. 2003; Folley & Park 2005; Hoffman
et al. 2004; Lauronen et al. 2004; Nettle 2001; in press;
Smalley et al. 2005). Many of these studies converge on a key
role for increased right-hemisphere activation in language function (Mohr et al. 2005). They also emphasize that understanding
psychosis requires analyses of its healthy analogue in components
of schizotypy, given the clear pathologies and fitness reductions
caused by psychosis itself, and the proposed “cliff-edged” form
of the balancing fitness function (Nesse 2005).
Finally, strong recent selection on language and cognition
coupled with antagonistic pleiotropy or linkage, and multilocus
overdominance, are not the only possible mechanisms for the
evolution and maintenance of psychosis in which selection
plays a central role. A non-exclusive model involves effects of
intragenomic conflict, mediated by sexual conflict or by
genomic imprinting in brain development (Badcock & Crespi
2006; Burt & Trivers 2006). The clearest evidence for genomicimprinting effects come from the oppositely imprinted disorders
Prader-Willi syndrome, which engenders high rates of psychosis
(Vogels et al. 2004), and Angelman syndrome, which shows a
high incidence of autism (Cohen et al. 2005; Peters et al.
2004). Genome scans also demonstrate strong imprinted-gene
effects in schizophrenia (Francks et al. 2003), bipolar disorder
(Kennedy et al. 2003), and autism (Badcock & Crespi 2006).
Genomic conflict may help maintain genetic variation via
continual strong selection for divergent optima, as in hostparasite conflicts mediated by major histocompatibility complex
(MHC) loci, the most polymorphic loci in the human genome.
Genomic imprinting effects also provide a persuasive hypothesis
for the paternal age effect on schizophrenia risk (Malaspina 2001;
Sipos et al. 2004), given that mutations during spermatogenesis
appear insufficient to explain such patterns (Farrer et al. 1992;
Reik et al. 1993; Tiemann-Boege et al. 2002). To the extent
that intragenomic conflict is involved in cognitive traits,
discussions of adaptation must focus at the level of genes, as
organism-level adaptive value can no longer be assumed (Burt &
Trivers 2006).
I am grateful to the Natural Sciences and Engineering Research Council
of Canada (NSERC) for research support and to A. Mooers, G. Cochran,
H. Harpending, and P. Nosil for helpful comments and discussion.
Why the adaptationist perspective must be
considered: The example of morbid jealousy
Judith A. Easton, Lucas D. Schipper, and Todd
K. Shackelford
Department of Psychology, Florida Atlantic University, Davie, FL 33314.
[email protected]
[email protected]
[email protected]
Abstract: We describe delusional disorder –jealous type (“morbid
jealousy”) with the adaptationist perspective used by Darwinian
psychiatrists and evolutionary psychologists to explain the relatively
common existence and continued prevalence of mental disorders. We
then apply the “harmful dysfunction” analysis to morbid jealousy,
including a discussion of this disorder as (1) an end on a continuum of
normal jealousy or (2) a discrete entity.
An evolutionary psychological approach to explaining the relatively common existence and continued prevalence of mental disorder historically has required explaining a disorder’s potential
adaptive benefits. As Keller & Miller (K&M) note, Darwinian
psychiatrists and evolutionary psychologists assume an adaptationist position, thus keeping natural selection at the etiologic
forefront. If it is theoretically possible and empirically verifiable
that mental disorder susceptibility alleles increased fitness in
some ancestral conditions, then a balancing-selection explanation
of the existence and prevalence of mental disorders may be
Delusional disorder – jealous type or “morbid jealousy” is a disorder that causes individuals to misinterpret everyday actions as
cues to a partner’s sexual infidelity. Constant accusations of infidelity, vigilant monitoring of a partner’s behavior, and restricting
a partner’s actions are typical of individuals diagnosed with
morbid jealousy (see the Diagnostic and Statistical Manual of
Mental Disorders, American Psychiatric Association [2000]; see
also, Kingham & Gordon 2004; Shepherd 1961; Vauhkonen
1968). The benefits and costs of morbid jealousy are well
documented (e.g., Buss 2000; Enoch & Trethowan 1979;
Kingham & Gordon 2004; Mowat 1966; Shepherd 1961). If
morbid jealousy is an extreme form of normal sexual jealousy,
it is reasonable to hypothesize that morbid jealousy may thwart
partner infidelity, perhaps more effectively than does normal
sexual jealousy, thereby increasing the fitness of ancestral
individuals with morbid jealousy. Whether the alleles associated
with the costs of morbid jealousy – such as decreased daily functioning, increased risk of mate defection, and increased susceptibility to other debilitating mental disorders – would be exactly
balanced through antagonistic pleiotropy by increases in the
fitness payoffs of the associated benefits is unknown. Despite
empirical challenges, an adaptationist perspective using balancing selection, specifically antagonistic pleiotropy, may explain
the relatively common existence and continued prevalence of
morbid jealousy and perhaps additional mental disorders.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
Wakefield (1999a; 2005) has argued that mental disorders can
only be classified as such when they are harmful dysfunctions. A
dysfunction is a failure of a mechanism to perform as it was
designed by natural selection. According to this definition, the
disorder cannot be the function of a naturally selected mechanism. Therefore, a dysfunction of jealousy mechanisms would
occur when they failed to motivate behaviors designed to
prevent a partner’s infidelity. Individuals diagnosed with
morbid jealousy do deploy behaviors that function to prevent
partner infidelity, even if the cues that activate the jealousy
mechanisms are imagined by the individual. Perhaps morbid
jealousy does not meet the dysfunction criterion and therefore
should not be considered a mental disorder.
Wakefield’s (1999a; 2005) harmful dysfunction analysis
specifies a second criterion that must be met for a mental disorder to be considered as such. The disorder must generate
harm, as defined by society. To conclude that morbid jealousy
is not a disorder without assessing the associated harm would
be a mistake, according to the harmful dysfunction analysis.
Lives are disrupted, including the lives of the morbidly jealous
individuals themselves as they constantly monitor their partner’s
behavior (e.g., Seeman 1979). Substantial stress is added to the
relationship as morbidly jealous individuals constantly accuse
their partner of infidelity (e.g., Vauhkonen 1968). Potential
rivals may be derogated or attacked, partners of the morbidly
jealous may be psychologically and physically abused, and sometimes this assault escalates to murder (e.g., Kingham & Gordon
2004; Mowat 1966; Shepherd 1961). Although morbid jealousy
is harmful, is it more harmful than normal sexual jealousy? In
fact, the greatest predictor of intimate partner homicide is
sexual jealousy (Daly & Wilson 1988). It is possible that morbidly
jealous individuals are more abusive toward their partners or are
more likely to murder them than are individuals who experience
normal sexual jealousy, but research has not investigated this
Morbid jealousy may be explained best not as a discrete
categorical mental disorder, but as a continuation of normal
sexual jealousy. Before this determination can be made,
however, several factors must be examined (J. C. Wakefield,
personal communication, March 20, 2006). First, we need to
determine whether the morbid jealousy tail of a normal curve
hides discrete points of jealousy disorders. For example, there
are many causes of low intelligence. However, a smooth
normal curve of intelligence would group these distinct causes
together and would hide the individual causes of low intelligence.
The same might be true of a normal sexual jealousy curve.
Examining individual cases of morbid jealousy and comparing
the symptoms and behaviors could help determine whether a
normal sexual jealousy curve is grouping together distinct
causes of morbid jealousy. If there are not multiple, distinct
cases of morbid jealousy, then it could be argued that morbid
jealousy is a continuation of normal sexual jealousy.
Second, we need to determine whether the morbid jealousy
end of a sexual jealousy curve is fitness enhancing. Previous
research has documented the adaptive benefits of normal
sexual jealousy; notably, that it may prevent partner infidelity
(e.g., Buss 2000). If morbid jealousy has similar adaptive benefits,
this might provide further evidence that it should be viewed as
part of a continuum of normal sexual jealousy.
Third, morbid jealousy may not be produced by a dysfunction
of jealousy mechanisms, but instead by a dysfunction of related
mechanisms. For example, individuals with morbid jealousy
may have dysfunctions in mate-retention mechanisms. If this is
the case, then morbid jealousy could not be considered continuous with normal sexual jealousy, as these related dysfunctions do
not occur with sexual jealousy. This third issue could be investigated by examining individuals diagnosed with morbid jealousy
to determine whether they have other, related dysfunctions.
Whether morbid jealousy is a discrete categorical mental
illness or part of a continuum of normal sexual jealousy
remains to be determined. We have discussed three research
questions that could help address this question. Investigation of
these questions through careful examination of individuals
with morbid jealousy may lead to clarification of delusional
disorder – jealous type, and may represent a model that could
be used to clarify other mental disorders. Additionally, this
clarification should lend support to continued use of the adaptationist approach and should provide a better understanding for
the continued prevalence of disorders.
Mutations, developmental instability,
and the Red Queen
Steven W. Gangestad and Ronald A. Yeo
Department of Psychology, University of New Mexico, Albuquerque,
NM 87111.
[email protected]
[email protected]
Abstract: We address two points. First, one must explain how different,
rare mutations ultimately lead to common psychopathological conditions.
The developmental instability model offers one solution. Second,
Keller & Miller (K&M) perhaps miss the major processes other than
variation fueled by rare deleterious mutations that account for
interesting genetic variation in psychopathology, particularly when
single alleles have non-negligible effects: Red Queen processes.
Keller & Miller (K&M) argue that much heritable variation in
psychopathological conditions is fueled by deleterious mutation,
rare at individual loci but ubiquitous in genomes. Ron Yeo,
colleagues, and I offered a similar view a decade ago (Gangestad
1997; Gangestad & Yeo 1997; Thoma et al. 2002; Yeo &
Gangestad 1993, 1998; Yeo et al. 1997, 1999), albeit less
broadly applied to neurodevelopmental disorders (e.g.,
schizophrenia, dyslexia, attention deficit disorder). K&M do an
excellent job of making this case.
Our quibbles pertain to details. We focus on two. First,
mutations at many loci produce phenotypic variants much
more common than individual mutations. Our model, curiously
not mentioned by K&M, may explain how they do so. Second,
K&M perhaps miss the most important alternative processes
accounting for interesting genetic variation in psychopathology,
particularly when single genes account for non-negligible
(.1%) variance.
The developmental instability model. Mutations at individual
loci are rare. Neurodevelopmental disorders are much more
common. A successful theory must explain how different
mutations can produce similar outcomes. Though K&M discuss
how different “upstream,” specific defects can have common
“downstream” effects (sect. 6.2), they do not present a
particularly compelling, specific model for how this happens.
We have suggested one route: developmental imprecision.
Microcircuitry of a computer chip must be manufactured in a
dust-free environment, for only then can its design be actualized. Dust that inadvertently becomes part of the chip can
affect the functioning of the circuitry in random ways, disrupting
design. Similarly, mutations and other developmental stresses
can act as “dust” in the environment in which epigenetic processes “manufacture” an organism’s phenotype, introducing
developmental instability and deviations from naturally selected
Neurodevelopmental errors may disrupt adaptive coordination
of a broad array of processes within developmental systems, particularly as their frequency increases. As disrupted development
may channel along particular paths, different perturbations (e.g.,
mutations) may ultimately have common outcomes. K&M argue
that more than half of all human protein-coding genes are
expressed in brain tissue and, hence, neural systems capture a
large amount of mutational variation. As genes that affect
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
neural development need not be expressed in brain cells per se,
however, mutations across an even larger proportion of the
genome may affect neurodevelopmental disorders. As this
theory expects, neurodevelopmental disorders covary with
markers of developmental disruption or instability (e.g., minor
physical anomalies and fluctuating asymmetry; see Yeo et al.
Disorder-specific features may arise because disruptions have
different outcomes depending on their timing or differentially
affect specific neural systems. Alternatively, disorders may be
affected by specific factors in addition to developmental instability (see Yeo et al. 1999).
Antagonistic coevolutionary processes. K&M contrast three
different kinds of evolutionary models of genetic variation:
mutation-selection balance, neutral variation, and balancing
selection – maintenance of variation because selection actually
favors multiple alleles (e.g., through heterozygote superiority,
spatiotemporal oscillations in selection, and frequencydependent selection). But a fourth possibility they do not
mention exists: rapid evolution at loci, particularly those
involved in antagonistic coevolutionary processes.
Species may antagonistically coevolve with other species (e.g.,
in predator-prey or host-pathogen roles). New adaptations in one
species evoke selection on the other to evolve counteradaptations. Antagonistic coevolution (or Red Queen processes; Van
Valen 1973) may persist through evolutionary time, resulting in
continual change in both species. Genetic conflicts of interest
between males and females, mothers and fetuses, or pairs of
cooperators may fuel persistent antagonistic coevolution of
genes within a species, as well (Rice & Holland 1997).
Red Queen processes produce predictable evolutionary outcomes: (1) Relatively rapid evolution: Because new adaptations
in a coevolving party change selection on the other party,
alleles at coevolving loci should be subject to new positive selection relatively frequently (e.g., Swanson et al. 2001; Wyckoff et al.
2000). (2) Interindividual variation: More often than most loci,
coevolving loci should be in states in which a new, favored
allele has not yet become near-fixed in the population and
coexist with alleles favored previously. (3) A non-negligible level
of maladaptation within populations: Neither conflicting party
is likely to adapt perfectly to the other party. The load of maladaptation will be carried disproportionately by individuals who
lack newly favored alleles.
A Red Queen process that might explain some variation in psychopathological conditions is maternal-fetal coevolution (see also
Gangestad & Yeo 1997). Fetal genes maximally benefit from a
greater flow of nutrients from the mother than the level maximizing maternal fitness (Trivers 1974). Maladaptive outcomes of
pregnancy (e.g., maternal hypertension, gestational diabetes)
may be by-products of ensuing antagonistic coevolution (Haig
1993). Maladaptive by-products could include disruption of
fetal epigenesis, resulting in neurodevelopmental disorder. In
this model, some fetal genes associated with poor development
(e.g., developmental instability) now once were not, for they
once did not coexist with current maternal genes that promote
counteradaptation. (The model is akin to spatially varying selection, but not as a form of balancing selection; cf. target article,
sect. 5.3.)
Some fetal genes involved in this conflict are imprinted –
expressed differently depending on parent-of-origin. Imprinting
is selected because fetal genes from fathers are also in conflict
with those obtained from mothers (Haig 2000). Among genes
that account for more than a negligible amount of variance in
psychopathology, imprinted genes (or other genes involved in
the imprinting process; e.g., genes that control imprinting; Burt
& Trivers 2000) may be over-represented, for example, on
schizophrenia, autism, or bipolar disorder (see Bah et al. 2004
[GRIK2]; Corradi et al. 2005 [GNAL]; DeLisi et al. 2002
[chromosome 22]; Francks et al. 2003 [chromosome 2]; Xu
et al. 2001 [CHRNA7]); on attention-deficit hyperactivity
disorder (ADHD) (see Borglum et al. 2003 [dopa decarboxylase];
Hawi et al. 2005 [9 genes putatively associated with ADHD];
Kirley et al. 2002 [TH]; Mill et al. 2004 [SNAP-25]). Only a
few percent of genes are imprinted (on the mouse genome, see
Luedi et al. 2005).
Findings in linkage studies have historically been fragile.
Cautious interpretation is advised. Nonetheless, even if most
variation in psychopathology is a result of rare mutation, some
single genes probably do have non-negligible effects. One evolutionary theory K&M do not mention points toward genes
involved in Red Queen processes.
Autism: Common, heritable, but not harmful
Morton Ann Gernsbacher,a Michelle Dawson,b
and Laurent Mottronb
Department of Psychology, University of Wisconsin–Madison, Madison, WI
53706; bPervasive Developmental Disorders Specialized Clinic, University of
Montreal, Hôpital Rivière-des-Prairies, Montreal QC H1E 1A4, Canada.
[email protected]
[email protected]
[email protected]
Abstract: We assert that one of the examples used by Keller & Miller
(K&M), namely, autism, is indeed common, and heritable, but we
question whether it is harmful. We provide a brief review of cognitive
science literature in which autistics perform superiorly to non-autistics
in perceptual, reasoning, and comprehension tasks; however, these
superiorities are often occluded and are instead described as
We appreciate Keller & Miller (K&M) grappling with the age-old
evolutionary paradox of why certain human phenotypes are so
common, so heritable, but so harmful. In their treatise, K&M
provide several examples of what they refer to as mental disorders, clumping together numerous phenomena, including
schizophrenia, bipolar disorder, depression, phobias, panic disorders, Tourette’s syndrome, obsessive-compulsive disorder,
low intelligence, anorexia, and autism. We – a cognitive scientist,
a research psychiatrist, and an autistic (who conducts cognitive
science research) – are most interested in K&M’s inclusion of
autism. Therefore, we restrict ourselves to that exemplar, agreeing that autism is common and heritable but questioning whether
autism is harmful.
Autism is definitely a common phenotype – even more
common than K&M report. Current prevalence estimates are
200 per 100,000 for DSM-IV (American Psychiatric Association
1994) defined “autistic disorder” and around 600 per 100,000
for the entire autism “spectrum” (Chakrabarti & Fombonne
2005). A rash of public attention has spotlighted what are considered dramatic recent increases in autism prevalence, but our
most reasoned logic suggests that the increases are due to purposely broadened diagnostic criteria, yoked with dramatically raised
public awareness and conscientiously improved case finding
(Gernsbacher et al. 2005). And when some lay spokespersons
mistakenly suggest that autism first appeared in society only in
the 1940s (Kennedy 2005), they are confusing the codification
of the phenotype with its onset (see Frith [1989] for a convincing,
albeit speculative, history of autism in society).
Autism is also a highly heritable phenotype, based on estimates
from twin studies and sibling-recurrence rates. However, the
existing heritability estimates warrant caution in interpretation.
The twin-based estimates are derived from only a handful of
studies, which are based on only a few handfuls of twins, and estimating sibling recurrence requires a reliable population prevalence rate.
But is autism a “harmful” phenotype? Primarily, K&M
employ an evolutionary connotation of harmful, namely,
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
lowered fitness (i.e., reduced fertility rates). Perhaps any
extreme phenotype will be less reproductively fit, be it the low
levels of intelligence that K&M include as an example or the
extremely high levels of intelligence found in adults identified
during adolescence by their academic precocity (Lubinski
et al. 2006). Certain cognitive phenotypes might also lead to
lowered fitness. The prolific inventor Nikola Tesla, who is
reported to have been celibate and whose life history reveals
numerous autistic traits, proclaimed:
I do not think there is any thrill that can go through the human heart
like that felt by the inventor as he sees some creation of the brain
unfolding to success. . . . Such emotions make a man forget food,
sleep, friends, love, everything. . . . I do not think you can name
many great inventions that have been made by married men. (Pickover
1999, p. 35)
K&M also verge into the more vernacular meaning of “harmful.”
They refer to mental disorders as “harmful dysfunctions” (sect.
1.2, para. 2), which are “disabling” and “debilitating” (sect. 1,
para. 2), which cause “human suffering” (sect. 1.1, para. 4), and
which are “disastrous to survival” (sect. 1.2, para. 6). K&M
view “mental disorders” such as autism as “glaring exceptions”
to the “awesome power of natural selection” (sect. 2, para. 1.).
However, whereas K&M assert that Darwinian psychiatrists
and evolutionary psychologists “often go to torturous lengths to
find hidden adaptive benefits” (sect. 1.1, para. 3), we assert that
cognitive scientists often go to torturous lengths to occlude
obvious adaptive benefits. The empirical literature is replete
with demonstrations of autistics’ superiority in numerous perceptual, reasoning, and comprehension tasks: Across a wide range of
age and measured intelligence, autistics perform significantly
better than non-autistics on block design, a prominent subtest
of Wechsler-type scales (Shah & Frith 1993); on embedded
figures tests, which require rapid visual identification of a
target figure amid a complex background (Shah & Frith 1983);
on recognition memory (Toichi et al. 2002); and on sentence
comprehension (Just et al. 2004); and autistics are more impervious than are non-autistics to memory distortions (Beversdorf
et al. 2000) and misleading prior context (Ropar & Mitchell
2002). Such superiorities are not isolated phenomena; some
theorists argue that such superiorities abound in autism
(Mottron et al. 2006).
Quite compellingly, each of these statistically significant demonstrations of autistic superiority is labeled by its authors as a
harmful dysfunction. Autistics’ superior block-design performance is labeled “weak central coherence,” symptomatic of dysfunctional “information processing in autism” (Shah & Frith
1993, p. 1351). Autistics’ superior performance on embedded
figures tests is considered “consistent with the cognitive-deficit
theory proposed by Hermelin and O’Connor (1970) . . . due to
a central deficiency in information processing” (Shah & Frith
1983, p. 618). Autistics’ superior recognition memory performance is attributed to deleteriously “enhanced attention to
shallow aspects of perceived materials” (Toichi et al. 2002,
p. 1424); their superior sentence comprehension is described
as being “less proficient at semantically and syntactically integrating the words of a sentence” (Just et al. 2004, p. 1816); their
superior imperviousness to memory distortions is explained by
“representations in the semantic network [that] may be associated in an aberrant manner” (Beversdorf et al. 2000, p. 8736);
and their superior resistance to misleading prior context is
attributed to their perception being “less conceptual” (Ropar &
Mitchell 2002, p. 652).
Disorders are defined by criteria that vary with cultural,
societal, and medical values. As K&M write:
Mental disorder categories may reflect a mix of historical convention,
diagnostic convenience, innate categorization biases in person perception, and common final pathways of partially overlapping yet distinct
dysfunctions. This suggests that the number of loci affecting a mental
disorder depends in large part on the way human minds categorize
behavioral symptoms. (target article, sect. 6.4, para. 6; emphasis in
We couldn’t agree more. As autistic Suzanne Shaw opines:
People say that in the world of the blind the one-eyed man is king, but I
think they are mistaken. In the world of the blind the one-eyed man
would be a freak, and his eye might even disable rather than enable
him. Eyes are wonderful things to be sure, but they are only useful
in a society that is built to require them. (
We would add that they are only useful in a society that is open to
appreciating them.
Heritable mental disorders: You can’t choose
your relatives, but it is they who may really
Ester I. Klimkeit and John L. Bradshaw
Department of Psychological Medicine, School of Psychology, Psychiatry and
Psychological Medicine, Monash University, Clayton, Victoria 3800, Australia.
[email protected]
[email protected]
Abstract: Keller & Miller (K&M) briefly mention and promptly dismiss
the idea that genes for harmful mental disorders may confer certain
advantages to affected individuals. However, the authors fail to
consider that the same genes (in low doses or reduced penetrance) may
be adaptive for relatives, and that this may in part explain why they are
retained in the gene pool.
While Keller & Miller (K&M) here rightly address ancestral
factors in developing their critique of the three evolutionary
models under their consideration, they only briefly consider
the idea that negative fitness effects from mental disorders
(e.g., schizophrenia, bipolar disorder) may be offset by other
potential benefits (e.g., creativity). They fail to consider that
genes for mental disorders may be retained, even selected
for, in the genome when they convey certain other advantages,
perhaps in different environments or contexts, not necessarily
in those directly affected by the disorder in question, but in
close relatives, when present in low doses or exhibiting
reduced penetrance. A model for this is the recent evidence
that genes associated with a homosexual orientation in males
(where retention and transmission in the genome are clearly
less likely in the affected male) are also associated with
increased fertility in close female relatives (Camperio-Ciani
et al. 2004). It is also well established that close relatives of
autistic individuals exhibit unexpectedly high frequencies of
choosing certain types of career where they excel, especially
those involving computation, information technology,
programming, and accountancy, where social and communicatory skills are at a reduced premium (see Baron-Cohen &
Belmonte 2005).
Similar factors may operate, additional to any possible
compensatory effect in the affected individuals themselves (see
Bradshaw [2001] for a review), with “song and dance” and ball
skills associated with Tourette’s syndrome, and excellence in
tasks involving attention to detail or general or personal care or
hygiene associated with obsessive-compulsive disorder. In
uncertain times or in the case of war or threats, the capacity
for suspicious and critical distrust (verging on the paranoic)
associated with schizophrenia, and the diffuse attentional focus
and impulsivity associated with attention deficit hyperactivity
disorder (ADHD), may be beneficial. Despite the denial by
K&M of a significant link between creativity and mood
disorders, there is indeed considerable evidence to support
such an effect – such as the study by Simeonova et al. (2005),
who found that children of adults with bipolar disorder had
higher scores than did controls on a measure of creativity (see
also Andreasen 2005). Although K&M correctly draw our
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
attention to the reduced reproductive fitness in schizophrenia,
they do not discuss the evidence that individuals with
ADHD tend to have more children than do controls (Weiss
et al. 1985).
K&M point out the arbitrariness between classification of what
is “normal” and what is “abnormal” because most mental disorders are extreme points along a continuum of symptom severity. From this viewpoint, the “normal” spectrum of symptoms
may have adaptive functions, such that normal depressed mood
may motivate avoidance of similar situations in the future.
Unlike somatic disorders, there is rarely an objective gold standard for diagnosis: Severities range widely and wax and wain
over time in a given individual; boundaries with normality are
fuzzy and shifting, depending often on society’s own changing
norms and expectations (Joan of Arc, a heroine in her day,
would probably nowadays be committed to an institution); and
boundaries between the disorders themselves are often fuzzy
and arbitrary with much overlap, irrespective of questions of
comorbidity, as if it all depends on which “joints” we happen to
choose with which to “carve” nature. Cultural factors may influence the prevalence and severity of mental disorders, because
cultural and societal tolerance for different behaviours vary
(McArdle 2004). Using the same cut-off scores on a behaviour
teacher rating scale for example, produces different rates of
hyperactivity in children of different countries (e.g., in Scotland,
4.5% of children are classified hyperactive, whereas, in Spain,
16% of children are classified hyperactive using the same criteria;
Gingerich et al. 1998).
Brown and Braithwaite (2004) found that, despite (and,
initially in fact probably because of) variability in both
groups, on average, fish introduced into high-predation
environments in a very few generations tended to become
more bold (showing a greater propensity to take risks and
greater exploratory behaviour) than did fish (from the same
original founder population) from low-predation sites. Presumably it is advantageous for fish in high-predation environments
to explore a new environment thoroughly to become aware of
escape routes and to ensure that no predators are present.
Thus, “personality” or temperamental traits appear to be
selected for, often in a very short time period, if they are advantageous in a particular environment. Generally, what is genetically transmitted is perhaps not so much a disorder per se, but
rather a particular kind of general personality bias which may
then predispose an individual to morbidity in a certain societal
context – or perceived excellence in another. There certainly
appears to be a connection between temperament, which is
considered to be an early precursor to personality traits
(Nigg et al. 2002), and psychiatric disorders, because difficulttemperament children are over-represented in psychiatric
populations (Maziade et al. 1990a). However, difficult temperament predicts the presence of psychiatric conditions in
preadolescence and adolescence only when family functioning
is also taken into account (Maziade et al. 1990b). Thus,
extreme temperament is not automatically equivalent to a
psychiatric disorder, and reflects the importance of considering
gene-environment interactions.
Inheritance of general personality factors may predispose an
individual to the risk of developing one or more of a range or
possible maladaptive (or even adaptive) behaviours, depending
on the individual’s environment (Legrand et al. 2005). Thus,
the same genes for externalising tendencies may be expressed differently under different environmental conditions, and predispose, on the one hand, both to antisocial behaviour and drug
and alcohol problems, and, on the other, to otherwise useful, if
risky or dangerous, occupations (e.g., test pilots, fire fighters).
Similarly, impulsive people may be well placed to take advantage
of unexpected opportunities, whereas others’ impulsive choice
may lead to drug addiction in which addicts affect their health
for the immediate rewards of the drug (Cardinal 2004;
Evenden 1999).
Are common, harmful, heritable mental
disorders common relative to other such
non-mental disorders, and does their
frequency require a special explanation?
Oliver Mayoa and Carolyn Leachb
CSIRO Livestock Industries, Adelaide, SA 5000, Australia; bSchool of
Molecular and Biomedical Science (Genetics), University of Adelaide,
Adelaide, SA 5005, Australia.
[email protected]
[email protected]
Abstract: Keller & Miller’s (K&M’s) conclusion appears to be correct;
namely, that common, harmful, heritable mental disorders are largely
maintained at present frequencies by mutation-selection balance at
many different loci. However, their “paradox” is questionable.
The “paradox,” which is largely set out in the first sentence
of Keller & Miller’s (K&M’s) abstract, has two elements: the
existence of common disorders agreed to be deleterious in
present-day environments and shown by the authors to reduce
reproductive performance (fitness) in many cases; and an effective mechanism for reduction of frequency of alleles predisposing
persons to deleterious traits. The reality of the paradox requires
consideration before assessment of K&M’s explanation for the
prevalence of disorders agreed to be common.
Mental disorders as a special category. We first address the
case of mental disorders (MDs), since they are considered at
length by K&M.
Table 1 lists a number of disorders well established as having
high population prevalence and substantial heritability. Heritability (h 2) is used as an indicator of genetic importance in
aetiology despite its well-known defects (mentioned by K&M),
because for common diseases there is no problem of h 2 being
We see immediately that traits other than MDs which certainly
reduce fitness, for example, type 1 diabetes and endometriosis,
are also at high frequencies which require explanation on
K&M’s argument. We also tentatively conclude that, however
special and important human mental abilities and disturbances
thereto may be to humans living in society, there is no reason
to separate them out for the purposes of assessing K&M’s
paradox. We shall therefore consider them separately only after
dealing with the two general propositions which constitute
K&M’s paradox.
The existence of common deleterious traits. The majority of
the traits shown in Table 1 are deleterious in present-day
societies, in terms of reduction in reproductive fitness. The
societies under discussion are for the most part characterized
by large numbers of unrelated people living in close proximity,
sustained by nutrition adequate to excessive for the low level
of physical effort required normally to gain a living in
employment and other activities which themselves differ
greatly from those of the first 24,000 generations at least of the
human species’ putative 25,000 generations of existence.
Some of these traits have increased substantially in frequency
in recent centuries. In some cases, environmental factors have
been identified as causal, for example, diet and exercise patterns
for ischaemic heart disease. Of type 1 diabetes, Hyttinen et al.
(2003) have written:
Type 1 diabetes among children 15 years has increased worldwide
during [recent decades]. In light of population genetics, the rate of
increase in the incidence . . . is too rapid to be caused by changes
in the population gene pool. Despite harmful effects of diabetesassociated alleles, they are common in many populations. Environmental risk factors may directly trigger the process leading to type 1
diabetes or may interact with diabetes susceptibility genes that
modify the penetrance. Heritability [in our study] was found to be
higher than that discovered before. If the increase in heritability is
real, it should be at least partly interpreted as a changed penetrance
of the diabetes susceptibility genes. (p. 1054)
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
Table 1 (Mayo & Leach). Prevalence and heritability of some common disorders in various
Prevalence %
Heritability %
Source (Reference)
Anorexia nervosa
Bipolar disorder
Congenital heart disease
Type 1 diabetes
Mild mental retardation
Ischaemic heart disease
Peptic ulcer
0.02 – 0.05
2 –5
5 –17
8 –10
15 –20
Smith 1975
Hyttinen et al. 2003
Smith 1975
Smith 1975
Smith 1975
Treloar et al. 2005
Smith 1975
Hence, such a change must have been environmentally
Other diseases, such as many infectious diseases, have
declined in incidence and prevalence in recent centuries. In virtually all cases where explanations have been obtained, environmental factors have been shown to be causal, even where genetic
susceptibility is implicated in causation of the disease. Environmental change has thus been important in changes in disease
incidence and prevalence upwards and downwards; it would be
of interest to know what diseases have remained unchanged in
incidence or prevalence in recent centuries, but rare Mendelian
recessive disorders could lie in this group. K&M present no
evidence on constancy of frequency of their target category:
“[M]ental disorders that are much more common than would
be expected from a single-gene mutation-selection balance;
roughly, this corresponds to mental disorders with lifetime prevalence rates above [0.05%] in reproductively aged adults” (sect.
1.3, para. 6). In the absence of evidence, one cannot reject the
simple hypothesis that some changes in the human environment
in the last thousand generations have contributed to an increase
in the frequency of disorders that are particularly deleterious in
large, organised societies not engaged in essential, risky, strenuous physical work. K&M address “ancestral neutrality” of causal
alleles of relevant genes in sections 3.3 and 4 but reject it because
of population-genetic considerations. Their argument concerning environmental change is brief and based on the implausibility
of large GE interactions and the rarity of strict neutrality
(sect. 4.2). Leaving aside discussion of such strict neutrality, we
simply note that very large GE interactions are inherent in
the increase in frequency of diabetes and various cancers; they
are not inherently implausible. Indeed, K&M accept in section
4.4 that GE interactions and nearly neutral variation could be
important in the very high incidence of depression, where
simple environmental causation of change in incidence has
indeed been invoked from time to time (e.g., Hibbeln 1998).
In the absence of evidence of constancy of frequency of MDs
over time, one cannot reject the hypothesis that environmental
factors have increased their frequency, making highly deleterious
alleles that were previously neutral, advantageous or slightly
deleterious. Neutrality is a second-order question until the
hypothesis stated has been tested.
Depletion of genetic variation by natural selection. Much of
K&M’s argument is based on the Fisherian concept that a
population will, other things being equal, increase in fitness at
a rate given by the additive genetic variance in fitness (see
Ewens 2001). On K&M’s argument, natural selection will
therefore deplete variance in fitness rapidly, apart from that
generated afresh by mutation. However, this conclusion
ignores two matters: First, the environment is never constant
and indeed may be viewed from the organism’s perspective as
constantly deteriorating (Fisher 1930/1999); and, second,
variance in fitness and associated metrics (e.g., heritability) are
not simply reduced rapidly to zero for “fitness traits” and
left as they are for “non-fitness traits” (see Bürger et al. 1989;
Keightley & Hill 1987; Mayo et al. 1990).
Traits which may be substantially influenced by the environment, such as all those listed in Table 1, should be considered
in the light of the cautions just expressed; environmental
change, potentially so much more rapid and far-reaching,
should always be evaluated before considering genetic change.
Indeed, Kirk et al. (2001) have drawn much the same conclusion
from a very thorough direct study of the heritability of fitness in
one human population.
Genetic contributions to causation of mental disorders. We
argue that K&M have not made their case in regard to either
the overall causation of MDs or the genetic evidence for
mutation-selection balance as the prime source of genetic
variance in MDs. However, as set out in the previous section,
we consider such balance as the major source of genetic
variance in many traits, among which could be MDs.
It is possible that those who hold the belief a priori that the
genetic basis of multifactorial traits is oligogenic may still find
this conclusion paradoxical in some way. However, we should
note that evidence from experimental organisms shows that
many traits are controlled by many – frequently hundreds – of
genes, and that there are scores of interactions among these
genes, even for simple quantitative traits in plants such as rice
(for discussion, see Mayo 2004). For truly complex traits such
as human mental development and function, influenced
perhaps by thousands of genes, as noted by K&M, it should
not be surprising that causation of variation is not oligogenic.
We note further that K&M have not made a convincing special
case for MDs as against other common familial diseases, and
conclude by quoting Bodmer (1999, p. 103), who has applied
the same arguments to common cancers:
[T]hese types of variants [rare variant alleles at many different loci] may
thus represent a major new facet of the study of multifactorial disease
inheritance, representing effects that lie between those of severe
clearly inherited susceptibilities and relatively common multifactorial
low-penetrance effects, such as are characterised by the many associations between polymorphic HLA variants and autoimmune diseases.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
The romance of balancing selection versus
the sober alternatives: Let the data rule
John J. McGrath
Queensland Centre for Mental Health Research, University of Queensland,
The Park Centre for Mental Health, Wacol, Queensland 4076, Australia.
[email protected]
Abstract: Schizophrenia has attracted more than its fair share of
evolutionary-based theories. The theories involving balancing selection
are based on the assumption that the incidence of schizophrenia is
invariant across time and place. Modern epidemiology allows us to
reject this dogmatic belief. Once variations in the genetic and
epidemiological landscape of schizophrenia are acknowledged, more
productive research models can be generated.
Keller & Miller (K&M) have provided a sobering analysis of
the competing models explaining the persistence of various
psychiatric disorders in the face of a reduction in fitness associated with these disorders. Their conclusions are both frustrating
and liberating. Rather than the romantic and quasi-heroic
notions associated with balancing selection, the evidence
suggests that the genetic contributions to these disorders are
probably due to the multi-generational accumulation of a wide
variety of mutations in a wide range of genes.
My comments focus on schizophrenia, which has attracted
more than its fair share of balancing-selection evolutionary
theories. The major evolutionary theories of schizophrenia have
been predicated on the belief that the incidence of schizophrenia
is invariant across time and place. This has influenced researchers to link susceptibility genes for schizophrenia to distant speciation events (e.g., Burns 2004; T. J. Crow 2000). In these theories,
symptoms or neurobiological correlates of schizophrenia are
interpreted as trade-offs for adaptations related to key
innovations of our species (e.g., language, theory of mind,
social intelligence, creativity, etc.). These theories are ingenious
and thought-provoking, but they also tend to be overly ambitious.
They build theoretical mountains out of empirical molehills.
K&M’s target article is timely because data from epidemiology
are now overturning some of the dogma surrounding schizophrenia (van Os et al. 2005). It is now clear that the central
notion underpinning balancing-selection theories of schizophrenia (i.e., that schizophrenia occurs with equal incidence
around the world) is wrong (McGrath et al. 2004). Not only are
there prominent variations in the incidence of schizophrenia
between sites, the risk of schizophrenia varies substantially
within populations. For example, subgroups that are at increased
risk of developing schizophrenia include men, first-generation
and second-generation migrants, those born and/or raised in
cities, the offspring of older fathers, and those born in winter
and spring.
When the detailed epidemiological landscape of schizophrenia
is appreciated, schizophrenia becomes less exceptional and more
like other disorders – its incidence varies across gradients in
place and time (McGrath 2006). The data show that romantic
and muddle-headed notions that schizophrenia respects human
rights (i.e., is an “egalitarian” disorder) are wrong (McGrath
2005). As K&M note, theories based on balancing selection
might be appealing for social and moral reasons. However,
when data no longer support the dogma, data must rule. Theories
based on polygenic mutation selection do not rely on a flat epidemiological profile of schizophrenia; hence, the recent advances in
epidemiology should not undermine K&M’s model.
The failure to appreciate variations in the incidence of
schizophrenia may have also hindered recognition of the
between-population variation in the prevalence of susceptibility
genes. This issue is relevant to the scenario outlined by K&M
in section 5.3 (“Temporal or spatial variability in fitness landscapes”). For example, while much research has focused on the
association between polymorphisms in the COMT gene and
risk for (and expression of) schizophrenia (Tunbridge et al.
2006), there has been less recognition that the prevalence of
polymorphisms in this gene varies considerably between populations (Palmatier et al. 1999). Similarly, the dystrobrevinbinding protein dysbindin is a much-cited candidate gene for
schizophrenia (Benson et al. 2004; Sullivan 2005) and perhaps
also psychosis associated with bipolar disorder (Raybould et al.
2005). However, little if any attention has been given to the
finding that this gene appears to be under strong positive selection in Europeans (perhaps related to its association with skin
pigmentation) (Voight et al. 2006). With respect to modern theories of gene by environment interactions leading to depression,
one of the leading candidate genes is the serotonin transporter
gene (SLC6A4) (Levinson 2006). The region containing this
gene also appears to be under positive selection in some, but
not all, of the populations assessed to date (see http://hgwen. Although the overall
genetic calculus suggests that balancing selection is unlikely to
account for the persistence of common heritable mental disorders, these examples may reflect more nuanced examples of
fine-grained temporal fluctuations caught in the “snapshot” of
evolution. It remains to be seen whether the factors influencing
the positive selection of these candidate genes can be determined. Relying on the surface-level adult phenotype to understand how candidate genes influence the matrix of brain
development can be a frustrating task.
Polygenic mutation-selection models also encourage us to shift
focus with respect to the category of observation in psychiatric
research. Rather than searching for an association between a
few genes and one particular group of psychiatric disorders, it
directs our scrutiny towards over-arching biological systems.
The question is no longer which gene and associated protein
causes the disease of interest, but which biological pathways in
which cells are most vulnerable to the cumulative mutational
It is important to recognize that the symptoms of psychiatric
disorders are emergent properties of highly complex and
robust systems. Evolutionary developmental biology (evo-devo)
reminds us of how evolution builds robust systems (Carroll
2005; Kitano 2004). Important properties become highly buffered over phylogenetic development. When robust biological
systems fail, occasionally we see catastrophic, cascading failures
(e.g., akin to the loss of consciousness associated with grand
mal epilepsy). However, in most circumstances, robust systems
call up other mechanisms to maintain output. The systems can
become fragile, though, in certain circumstances. In this
respect, the broad sweep of mutations proposed in the K&M
model may be sufficient to unmask fragilities, leading to the
neurological equivalent of graceful degradation (a term from
computer programming used to describe the ability of software
to continue operating with reduced function rather than
“crash”) (Bentley 2004).
Can these notions provide directions for future schizophrenia
research? Can we find the appropriate biological category of
observation that will enable us to interpret the data that polygenic mutation-selection theories predict will be found? Evidence from genomics and proteomics based on post-mortem
brain tissue is already providing candidate biological systems
for closer scrutiny. For example, many genetic and environmental factors can subtly impact on the cross-talk between pathways involved in apoptosis and synaptic plasticity (Catts & Catts
2000; Glantz et al. 2006), or subtly reduce the efficiency of
mitochondrial trafficking (which is increasingly being recognized
as important for neuronal functioning) (Reynolds et al. 2004).
When research finds the right category of observation, order
may be found amongst disorder.
If common, heritable mental disorders are underpinned by
multi-generational cumulative sweeps of ever-changing mutations,
then the search for susceptibility genes will be a substantial
challenge. In the context of schizophrenia, the complexity of
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
this model complements that growing awareness that the incidence of schizophrenia has prominent variations across time
and place. However, I argue that variations in the incidence of
schizophrenia should be seen as valuable opportunities to generate and test novel candidate exposures. These exposures operate
against a diverse and ever-changing backdrop of susceptibility
genes. Linking these genes into biological systems may provide
us with clues about how the environment can help optimize
brain development and reduce system fragility. Having a sober
and data-based understanding of the complexity of genetic and
environmental risk factors underpinning common mental disorder is a crucial step in helping neuroscience reverse-engineer
the complex systems underpinning brain development.
The author was supported by Queensland Smithsonian Fellowship and
the Stanley Medical Research Institute.
Reconciling the mutation-selection balance
model with the schizotypy-creativity
Daniel Nettle
Evolution and Behaviour Research Group, Division of Psychology,
University of Newcastle, Newcastle NE2 4HH, United Kingdom.
[email protected]
Abstract: Keller & Miller (K&M) make a persuasive case for the role of
mutation-selection balance in the persistence of such disorders as
schizophrenia. However, there is evidence relating illness liability to
creativity, which seems to imply balancing selection. I argue for a hybrid
position, where schizotypal personality traits can have fitness advantages
or disadvantages, with mutational load and neurodevelopmental
conditions determining which outcome is observed.
The authors of the target article make an elegant and extremely
persuasive case for the role of polygenic mutation-selection
balance in the persistence of traits that impair mental functioning
in humans. Keller & Miller’s (K&M’s) article is very important
for a number of reasons. First, within psychiatric genetics,
there has been an implicit assumption that genes predisposing
people to mental disorders will be easy to identify in the way
that has been true of those involved in Mendelian disorders.
The mutation-selection balance model shows persuasively why
finding a few genes of major effect that have their effects in all
populations is a naı̈ve expectation. Second, within evolutionary
psychology, there has been excessive attachment to the idea
that traits of importance to fitness should show no heritable variation. This has led to some odd accounts which sought to imply
that mental disorders usually or always represented the proper
functioning of universal adaptations, and only seemed to be
maladaptive because of the modern environment, or medical
stigmatisation (see Nettle 2004). The target article rightly introduces evolutionary psychology to the fact, now well understood
by evolutionary biologists, that even traits under strong selection
can maintain abundant genetic variation, if the mutational target
size is sufficient.
The authors rightly stress that strong empirical evidence for a
mutational load account for serious disorders such as schizophrenia comes from such findings as paternal age effects and
inbreeding depression. There are also non-heritable risk
factors. The authors allude to brain trauma as one such risk,
but neglect to point out that low birth weight, winter birth,
maternal infection during pregnancy, urban residence, and
exposure to house cats are all risk factors. Though these are
non-genetic effects, they are generally supportive of K&M’s
position, because they imply, more generally, that any factor
impairing neurodevelopment – be it deleterious mutations or
early life stressors – is a potential risk factor for schizophrenia.
This is consistent with the logic of the watershed model, where
failures of attention and social cognition would be the result of
many types of upstream instability, a point further reinforced
by the finding that schizophrenia patients have increased levels
of physical asymmetry compared to controls (Yeo et al. 1999).
For these reasons and more, the authors’ dismissal of balancing-selection models in favour of mutation-selection models
seems justified. The most commonly discussed balancing-selection model for serious mental disorders – that of a relationship
to creativity – is given short shrift. However, the empirical evidence for such a linkage is really quite strong, and much stronger
than the authors imply. The crucial finding is not that rates of
mental illness are higher in creative groups (though they are;
cf. Andreasen 1987; Ludwig 1995). Rather, the key finding is
that there are measurable cognitive affinities between those
successful in the creative professions and those diagnosed with
serious mental illness (Nettle, in press; Nowakowska et al. 2004;
Schuldberg 2000; Woody & Claridge 1977), whether or not the
creative individuals show any symptoms of psychopathology.
Thus, it seems likely that there is a shared endophenotype,
including broad attentional sampling and an inclusive style of
mental association, which is disproportionately found in both
healthy creative individuals and psychotic patients.
Most commentators have taken this as evidence for balancingselection effects (Nettle 2001). How then do we square such
evidence with K&M’s convincing case for the importance of
mutational load? I suspect that the answer lies in how mutational
load (and developmental instability in general) interacts with a
schizotypal cognitive style. Clegg and I have suggested that
where this cognitive style is coupled with stable neurodevelopment (low mutational load and benign environment), the result
is healthy creativity, whereas where it is coupled with genetic
and/or environmental instability, the result is serious mental
illness, with all its concomitant neurodevelopmental delays,
impairments, and fitness reductions (Nettle & Clegg 2006).
If this model is right, then the alleles leading to schizotypal
cognitive style are probably nearly neutral, on average, and
therefore a significant amount of variation is maintained. Thus,
the model squares the evidence for a link between schizotypy
and creativity with the evidence for mutational load and environmental effects on schizophrenia. The case for this model would
be much strengthened if there were evidence for increased
fitness associated with creativity. Note that the increased fitness
need not be restricted to siblings of patients, as long as it was
found in some individuals with a schizotypal cognitive style.
Clegg and I showed (building on ideas in Miller’s earlier work)
that artists and poets, who show some schizotypal traits, have
increased numbers of sexual partners relative to controls
(Nettle & Clegg 2006). Therefore, I feel that a hybrid model,
with balancing effects of schizotypal cognition, but mutational
load and neurodevelopment determining whether it is the negative or positive sequelae that develop, is very plausible.
Mental disorders are not a homogeneous
Joseph Polimeni
Department of Psychiatry, University of Manitoba, Winnipeg, Manitoba,
R3E 3N4, Canada.
[email protected]
Abstract: The only commonality between the various psychiatric
disorders is that they reflect contemporary problematic behaviors.
Some psychiatric disorders have a substantial genetic component,
whereas others are essentially shaped by prevailing environmental
factors. Because psychiatric ailments are so heterogeneous, any
universal explanation of mental illness is not likely to have any clinical
or theoretical utility.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
The target article by Keller & Miller (K&M) reviews potential
evolutionary genetic explanations for psychiatric disorders. This
is a welcome topic because the nascent field of adaptive
genetic variation may be applicable to certain psychiatric
conditions. However, some of the authors’ ideas are founded
on contentious suppositions.
K&M fail to distinguish between psychiatric conditions that
are genetically based and those that are mostly precipitated by
unique cultural factors. They mistakenly clump all psychiatric
disorders as being both heritable and reflecting brain “malfunction” (see, e.g., sect. 1.2). For the majority of psychiatric ailments,
the environment, and not heredity, is the primary determinant.
Eating disorders, for example, were rare until food became plentiful in Western societies. Borderline personality disorder is
uncommon in the absence of childhood neglect or abuse
(Bandelow et al. 2005; Zanarini 1997). Depressive episodes are
almost always precipitated by a loss of attachment or diminishment in status. It is acknowledged that genetic factors probably
alter vulnerabilities to various psychiatric ailments, but no one
knows how many genotypes could be relevant for any given
emotional problem. It is possible that only three classic psychiatric disorders (bipolar disorder, schizophrenia, and obsessivecompulsive disorder) possess an appreciable genetic component.
K&M’s greatest leap of faith is their non-categorical approach
to mental disorders – a necessary perspective for their polygenic
mutation-selection balance idea to work. Psychiatric disorders
are simply not as homogeneous as K&M assert. Schizophrenia,
bipolar disorder, obsessive-compulsive disorder, brain trauma,
and Down’s syndrome, for example, demonstrate remarkable
similarities within each diagnostic group, yet are all distinctly
different from one another. It is almost indisputable that schizophrenia and bipolar disorder tend to follow separate family genealogies (Cardno & Murray 2003; Lapierre 1994; Loranger 1981;
Maier et al. 1993; Potash et al. 2003; Smoller & Finn 2003;
Somnath et al. 2002; Taylor et al. 2002; Torrey et al. 1994).
Improved diagnostic methodologies have shown that cerebral
ventricular enlargement occurs in patients with schizophrenia
and their unaffected kin, but not in bipolar disorder (McDonald
Schizophrenia and bipolar disorder may sometimes be difficult
to distinguish clinically, but this in no way seriously invalidates
each respective diagnosis (there are several possible mundane
explanations for this apparent overlap).
K&M’s polygenic mutation-selection balance hypothesis
appears to be the genetic counterpart of Randall’s “misconnections” model explaining schizophrenia (Randall 1983). First
published in 1983, Randall proposed that novel neural pathways
could be established randomly, resulting in “supernormal
connections” or “misconnections,” resulting in various mental
disorders such as schizophrenia. A “biological trial and error of
connection would produce a range of behavioral variants”
(Randall 1998, p. 144). The inherent weakness of both Randall’s
proposal and the polygenic mutation-selection hypothesis is that
neither idea can comprehensively explain the characteristic
specificity of the various psychiatric disorders.
Contrary to popular belief, psychosis does not reflect random
scrambled thoughts. Psychotic delusions and hallucinations have
their own intrinsic patterns. For example, Polimeni and Reiss
(2004) have clinically observed that the vast majority of psychotic
delusions are either paranoid or spiritual in nature (consistent
with our shamanism-group selection hypothesis for schizophrenia). Remarkably, paranoid delusions rarely reflect veritable
dangers in proximity but usually involve suspicions outside
immediate surroundings. Although difficult to prove, it appears
that severe medical illness (without delirium) tends to mitigate
schizophrenia, mania, melancholia, and obsessive-compulsive
symptoms – as if these psychiatric ailments are fixed-action
behaviors disrupted in medically compromised individuals. The
presence of hallucinations and paranoid delusions (rarely spiritual delusions) in brain trauma, delirium, or dementia does not
necessarily mean that these symptoms reflect disease states in
every context. Vomiting, for example, is adaptive when expelling
foreign toxins but reflects a disease state when caused by a brain
K&M imply that psychiatric conditions manifest themselves
with the same severity in traditional societies; however, the
anthropological literature does not support this view. Most contemporary psychiatric disorders, including post-traumatic stress
disorder, agoraphobia, or classic obsessive-compulsive disorder,
may be rare in traditional societies (Polimeni et al. 2005).
Although schizophrenia-like symptoms are commonly described
(typically in shamans), they are less incapacitating. Suicide is not
infrequent, but often observed in the immediate aftermath of lost
K&M are correct that historical psychiatric epidemiology
could well pose formidable problems for evolutionary theories
and that this problem has generally been ignored. However,
very little is known about the fecundity of various psychiatric
conditions before the 1950s. Even less is known during the
transition from traditional societies to modern life over the last
few thousand years. Michel Foucault’s suggestion that mental
illness was socially unimportant before eighteenth-century
industrialization is not farfetched (Foucault 1965/1988).
K&M suggest that evolutionary theorists are often too quick to
invoke balancing selection because it is a convenient mechanism
to explain the adaptive qualities of mental illness; however, this
does not necessarily negate their position. In fact, when I first
embarked on evolutionary research, I naively examined group
selection with the mistaken belief that I was avoiding balancedselection arguments. In the end, I came to realize that the
necessary prerequisites for balanced selection are particularly
prominent in bipolar disorder, schizophrenia, and obsessivecompulsive disorder. For example, there is varying evidence for
heterozygote advantage and assortative mating in each of these
In an attempt to categorize the mechanisms of genetic variation, K&M may have oversimplified the dynamics of nature.
For example, it is possible that multiple mechanisms may
underlie any given type of variation, piggybacking or priming
the other (i.e., neutral drift þ heterozygote advantage þ
assortative mating). In other words, the machinations of genetic
variation are still very much a black box (Mousseau et al. 2000).
Notwithstanding the glib and amorphous idea of spandrels,
almost every physical attribute pertaining to a living organism
has been adaptive for at least a few generations. Even the circuitous route of the laryngeal nerve is no exception. Although there
may have been evolutionary design constraints, the entire “illogical” length of the laryngeal nerve is an adaptation! In humans,
universal genetic-based behaviors such as anger, jealousy,
humor, and attachment are no different. Although mental ailments may seem as imperfect as the circuitous laryngeal nerve,
certain psychiatric conditions plausibly make adaptive sense in
the primal world of traditional societies. Both disease and evolutionary models of psychiatric illness possess their own loose
ends. K&M effectively demonstrate that neither camp can be
dismissed outright.
Mental disorders, evolution, and inclusive
Antonio Pretia and Paola Miottob
Department of Psychology, University of Cagliari, 09123 Cagliari, Italy;
Department of Mental Health, Unità Locale Socio-Sanitaria (ULSS) 7, 31015
Conegliano, Italy.
[email protected]
[email protected]
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
Abstract: Grouping severe mental disorders into a global category is
likely to lead to a “theory of everything” which forcefully explains
everything and nothing. Speculation even at the phenotypic level of the
single disorder cannot be fruitful, unless specific and testable models
are proposed. Inclusive fitness must be incorporated in such models.
Speculation about the evolutionary origin of mental disorders is a
peaceful diversion from the challenge of attempting to treat
people diagnosed as having severe mental disorders. Keller &
Miller (K&M) efficiently dismantle this diversionary toy. What
remains of Darwinian psychiatry after their compelling criticism
is an approximate model of how the genes leading to severe
mental disorders had not been phased out from the general
population by natural selection. However, as K&M seem well
aware, “severe mental disorders” are very much influenced
“by the cultural and inherent person-perception biases . . . and
the categorization demands of legal, medical, and research
systems” (sect. 8, para. 5). So, what is the problem? Their
designation of severe mental disorders as a global category
predisposes to a “theory of everything” which explains everything
and nothing, the most common fault of evolutionary
We introduce as examples several specific and testable models
at the phenotypic level of the single disorder. First, let us
consider sociopathy: People with sociopathy cheat in reciprocity
games – they obtain resources and seldom return them. According to neo-Darwinian formulations, they challenge the cooperative subject (Axelrod & Hamilton 1981; Trivers 1971). Groups
of altruists who cooperate and are unable to detect cheaters are
easily exploited, leading to their extinction. However, groups of
altruists who are able to detect cheaters and discriminate in
their cooperative moves will protect their resources (Fehr &
Fischbacher 2003). In this regard, sociopathic individuals may
improve the altruists’ fitness overall. Since the cognitive abilities
leading to the cheater’s detection are likely to be useful in all
kinds of cooperative exchange (Stevens & Hauser 2004),
groups of discriminative cooperators will outcompete over
groups of non-discriminative cooperators. Sociopathy thus can
select for individuals who are more able to detect cheating. Its
permanence, thus, is of a parasitic kind: The hosts tolerate
some sociopaths in their environment because the sociopaths
continuously challenge the hosts’ cognitive abilities, as parasites
resident in our skin stimulate the immune system and act as a
restraint against more virulent invaders.
Patients with schizophrenia often need permanent help: They
deplete resources because they are unable to return anything.
Let us assume that alleles for schizophrenia spring up in two
clusters of families: the families that always help their kin, and
the others who never help their kin. Some of the helped kin
will arrive at reproducing their own susceptibility alleles,
whereas those without help will become extinct. Since helping
kin is a trait likely to favour inclusive fitness of the helper
(Hamilton 1964), after some generations the alleles for schizophrenia will spread in the population, on account of their
hitch-hiking on the helping trait, without adding any hidden
benefit to humankind: They fixate in the universal gene pool
because they are occasionally linked with some trait inherent to
our more general adaptive outfit. Indeed, incidence of
schizophrenia does not greatly vary across sites, confirming that
the disorder is rooted in our common genetic heritage. Cultures
more likely to display a helping attitude towards affected people
(such as some Indian enclaves), however, are also more likely to
have incidence rates that are higher than the mean (Jablensky
et al. 1992), together with a more favourable outcome (Leff
et al. 1992), leading in the long term to lower-than-average
prevalence rates (Saha et al. 2005).
Female patients with severe anorexia nervosa, by maintaining
their body weight below the threshold for ovulation, exclude
themselves from reproduction; however, they often behave in a
supportive way towards their kin, cooking for them what they
themselves are unwilling to eat. This behaviour is assimilated to
the “helping at the nest” behaviour described in the wild field
and observed to improve kin’s reproductive success (Arnold &
Owens 1998). Indeed, what matters for the permanence of a
gene set is persistence of that set in the pool, generation after
generation: Humankind, for example, is thought to carry on the
mitochondria of seven ancestral females (Sykes 2001). A
“helping at the nest” hypothesis for anorexia nervosa is a testable
one, different from the scenario of the “gene for facing famine”
(cf. Guisinger 2003).
It is difficult to understand K&M’s polygenic mutationselection balance model: They consider balancing selection as a
dynamics whereby two or more alternative alleles are maintained
because their net fitness effects balance each other out, so the
alleles are not lost by chance or genetic drift (Wilson 1998). A
polygenic mutation-selection balance, therefore, is a model
whereby a conditional balance is achieved: A mutation produces
a decrease in fitness only given a concurrent genetic environment; in the absence of such an environment, the mutation is
neutral. Past studies found that the genetic-controlled conditions
increasing the risk of obstetric complications are also associated
with a higher risk of schizophrenia, for example, in the case of
Rh incompatibility (Hollister et al. 1996). Conversely, obstetric
complications tend to recur within families, clustering in families
that also show a higher representation of subjects diagnosed with
schizophrenia (Walshe et al. 2005). Some time ago, we suggested
that the genetics of schizophrenia might be explained in part by
the genetics of the conditions increasing the risk of obstetric
complications (Preti et al. 1998). Foetal brain anoxia likely to
result from obstetric complications generally leads to death or
to severe motor impairment: We therefore hypothesized that
some brain-protecting gene would be necessary to balance the
brain-damaging impact of obstetric complications (Preti &
Miotto 2005). Whenever obstetric complications occur, schizophrenia develops only in the presence of the protecting gene.
A subgroup of offspring of patients diagnosed with schizophrenia
was found to bear a statistically reduced risk of developing
schizophrenia in adulthood, indeed, as if they were carrying
some protective gene (Gottesman & Erlenmeyer-Kimling
2001). Moreover, some studies reported a higher prevalence of
successful creative abilities in the mathematical, visual, and
spatial domains among the relatives of patients diagnosed with
psychosis (Karlsson 1999). Anomalous lateralization (Dragovic &
Hammond 2005) could be the link between schizophrenia
and superior mathematical ability. In the absence of severe
obstetric complications, protective, “left brain” genes would
favour creative abilities in the visual and spatial domains; in the
presence of such complications, schizophrenia would develop.
Is this a possible example of the kind of polygenic balance
K&M have in mind?
Behavioural ecology as a basic science for
evolutionary psychiatry
John S. Price
Odintune Place, Plumpton, East Sussex, BN7 3AN, United Kingdom.
[email protected]
Abstract: To the evolutionarily oriented clinical psychiatrist, the
discipline of behavioural ecology is a fertile basic science. Human
psychology discusses variation in terms of means, standard deviations,
heritabilities, and so on, but behavioural ecology deals with mutually
incompatible alternative behavioural strategies, the heritable variation
being maintained by negative frequency-dependent selection. I suggest
that behavioural ecology should be included in the interdisciplinary
dialogue recommended by Keller & Miller (K&M).
Keller & Miller (K&M) say that the heritability of mental disorders presents a problem for explanations in terms of function,
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
because alleles conferring function should increase in frequency
to fixation. They claim the only satisfactory explanation for the
high rate of mental disability is that mental disorders are
caused by harmful mutations on hundreds of genes. This argument restricts evolutionary psychiatry to disorders which have
zero heritability. However, the authors might have underestimated the prevalence and robustness of negative frequencydependent selection.
I would like to draw the authors’ attention to the discipline of
behavioural ecology, which is the study of behaviour in relation
to its function (Krebs & Davies 1993). Behavioural ecology
could be said to be the basic discipline of evolutionary psychiatry. It is concerned with strategy sets, which are sets of alternative strategies for dealing with problems. For instance, the cold
weather and reduced food supply of winter present a problem
to many species. Sometimes all members of a species deal
with the problem in the same way, but sometimes there are
alternative strategies for dealing with the problem. Many bird
species migrate; in some of these all the individuals migrate,
in other species only a proportion migrate and the rest stay
where they are. In very cold winters the rewards of migrating
are greater than staying, but in mild winters the rewards of
staying are greater. It is not of great concern to behavioural
ecologists just how the decision, to stay or migrate, is made.
It could be entirely genetic, so that a “staying” allele (or
group of alleles) is competing with a “migrating” allele. Or it
could be entirely environmental; for instance, it is thought
that robins compete for territories in the autumn, and those
birds who win territories stay and those who fail to win territories migrate. Probably for most partially migrating species
the decision-making mechanism is not known. Similar considerations may apply to partially hibernating species of rodents;
the territory owners stay awake, and those who do not have
territories go to sleep.
Coming closer to evolutionary psychiatry, let us consider the
case of pairwise contests. A rival for mates or other resources
poses a problem for the individuals of most species, and
various strategies have evolved to deal with it (Boone 1992;
Crowley 2003). Most species have evolved the alternative strategies of escalation and de-escalation. In territorial species, you
either fight or run away. In group living species, there is an alternative – appeasement – which enables one to continue living in
the group, albeit at a lower social rank. Each strategy has costs
and benefits. Behavioural ecologists such as Maynard Smith
have studied the conditions under which alternative strategies
could survive (Maynard Smith 1982; Parker 1984; Reichert
1998). Calling the escalating strategy the “hawk” strategy and
the de-escalating strategy the “dove” strategy, they concluded
that under certain conditions a mixture of hawk and dove is an
evolutionarily stable strategy (ESS) in that it cannot be infiltrated
and replaced by any other strategy, and in particular it cannot be
replaced by a pure hawk or a pure dove strategy. Thus, variation
in fighting behaviour is an ESS. The variation is maintained by
negative frequency-dependent selection, because in a world of
hawks it pays to be a dove, and in a world of doves it pays to
be a hawk. It does not matter whether the choice between
hawk and dove is genetically or environmentally determined.
Nor does it matter whether the choice is a “once and for all”
affair, such that type of parenting or some other variable made
an individual hawk or dove, or whether each individual has the
capacity to deploy both hawk and dove strategies, the choice
depending perhaps on environmental cues or possibly on a
random basis.
Both hawk and dove strategies have costs and benefits. We
think on the whole that the costs of being a hawk tend to take
the individual to the casualty department, whereas the costs of
being a dove take him or her to the psychiatric clinic (Price
et al. 2004). In other words, the costs of being a dove represent
some of the “mental disorder susceptibility alleles” of K&M.
And, since the choice between hawk and dove can be either
environmental or genetic, the problem of the partial heritability
of the mental disorders does not affect the argument.
Another issue is dispersal. Many species have both a maintenance phenotype, which is adapted to the natal territory, and a
dispersal phenotype, which is adapted to occupying new habitats
(Geist 1989). As each phenotype becomes rarer, its fitness
increases, so both are maintained by negative frequencydependent selection. During hominid evolution, rapid dispersal
must have been advantageous, as receding ice sheets left new
land available for occupation. However, human groups tend to
be united by a common belief system which differs from the
belief system of all the groups they are competing with. In
order to facilitate dispersal, it may have been advantageous for
an individual to undergo a change of belief system and to
convert some of the group to the new belief system, and to
take them off to a “promised land” (Price & Stevens 1999;
Stevens & Price 2000b). Thus, two dispersal phenotypes may
have evolved: One is the schizotype who has the capacity to
undergo a change of belief system, and one is the suggestible
or dissociative person who has the capacity to be converted
from the belief system with which he or she was indoctrinated
during childhood and to adopt the new belief system of a
prophet or cult leader. The fitness costs of both these dispersal
phenotypes would be grievous if dispersal was unsuccessful,
but the benefits of successful dispersal might also be very
great, leading to an adaptive radiation in a new habitat. In 45
years of psychiatric practice, I have seen many patients labelled
schizophrenic who in different circumstances might have
become effective cult leaders
I am indebted to the late Michael Chance and members of his
“Birmingham Group,” to Russell Gardner Jr., editor of the ASCAP
(Across Species Comparisons and Psychopathology) Newsletter, and to
members of the ASCAP Society for discussion of these ideas.
Bipolar disorder evolved as an adaptation to
severe climate
Julia A. Sherman
6302 Mineral Pt. Road, #303, Madison, WI 53705.
[email protected]
Abstract: Keller & Miller (K&M) assert that mental disorders could not
have evolved as adaptations, but they fail to make their case against
the theory of the evolutionary origin of bipolar disorder that I have
proposed (Sherman 2001). Such an idea may be unorthodox, but it has
considerable explanatory power and heuristic value.
In a previous publication (Sherman 2001), I proposed the theory
that circular bipolar disorder (major depression with hypomania
or mania) evolved as an adaptation to long severe winters and
short summers, which implicates the circadian clock in its pathophysiology. This idea is inferred from theorizing and data about
the seasonal effects of light and the small, but statistically significant, correlation between bipolar disorder and a pyknic, coldadapted build. The hypothesis is consistent with recent research:
Light therapy is as effective as antidepressant medications for
seasonal depression, and three studies have demonstrated similar
effectiveness for nonseasonal, major depression (Golden et al.
2005; Rosenthal 2006). Individuals with seasonal affective disorder,
compared with healthy volunteers, generate a biological signal of
change of season that is similar to the signal that mammals use to
regulate seasonal changes in their behavior (Wehr et al. 2001).
Other research genetically implicates the circadian clock in the
pathophysiology of bipolar disorder (Yin et al. 2006).
I have suggested that circular bipolar disorder evolved among
a small homogeneous population who lived in the northern
temperate zone of the Old World during the ice ages. Bipolar
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
depressive behaviors include lethargy, social withdrawal, and loss
of interest in food and sex. These behaviors conserved energy and
reduced social conflict within small groups holed up together for
the long, dark winter. Symptomatic cognitive impairment makes
sense in this context: winter was a time to endure, not a time for
thinking or action. During the short season of fair weather, the
active, outgoing, goal-oriented behaviors of hypomania helped
the group make up for lost time, accomplishing tasks necessary
for survival. Hypomania is a positive mood and energy state
associated with productive achievement, and bipolar disorder is
correlated with intelligence and upper socioeconomic level
(Goodwin & Jamison 1990), factors which Keller & Miller
(K&M) failed to mention. In addition to the 1% of the population
with severe symptoms, the authors’ major focus, another 5%
show milder forms of bipolar disorder with more positive associated behaviors (Jamison 2005).
Women in the ancestral bipolar disorder group may have had
a reproductive advantage because a certain level of body fat is
essential for ovulation, successful pregnancy, and breastfeeding. Reviews of the literature (Anastasi & Foley 1949;
Eysenck 1947, 1953; Sherman 2001) agree with Kretschmer’s
(1970) early-twentieth-century observation of a relationship
between circular bipolar disorder and a compact, pyknic build
with cold-adapted fat accumulation (also see Eiben et al. 1980;
Tóth et al. 2002). Women’s activity level and eating behavior
were adapted to their reproductive status, and eating heartily
during fair weather to store fat for winter is mirrored in what is
now called summer depression. Depression is at least twice as
common among women, and recent data show that it may be
more heritable among women (Kendler et al. 2006).
In contrast, mania, which is more frequent and intense among
males, may have evolved as a response to emergencies. During
mania, individuals have a superhuman ability to go without felt
need for sleep or food. They think fast and move fast, which are
characteristics that aid individual and group survival. Impulsive,
frequent sexual activity is also common during mania, but the
authors fail to consider the possible distortion of data because of
children born out of wedlock. In any case, bipolar fitness is much
higher than the level of fitness associated with schizophrenia.
Skepticism that circular bipolar disorder evolved as an
adaptation is understandable when it is taken out of context and
considered in its most extreme form. Psychosis is only sometimes
associated with bipolar disorder, and the minority who experience
a psychotic episode usually recover completely from the episode.
In contrast, a schizophrenic diagnosis implies psychosis, and about
one-third never recover. Another challenge to circular bipolar disorder as an adaptation is the current high rate of suicide associated
with the diagnosis. However, the rate of suicide now may not be
relevant to a time when individuals struggled merely to live and
when social conditions were more favorable, in the sense that,
because their behaviors were modal, individuals with bipolar
disorder were not stigmatized, victimized, or forced to function
at a high level during winter months.
The authors assert that bipolar disorder could not have evolved
as an adaptation because it increases with negative factors, such
as paternal age (no evidence cited), brain injury, and inbreeding.
Adaptations, however, are necessarily positive only for the selective environmental pressures that produce their evolution.
Although the authors discuss genes by environment interaction,
they fail to explore adequately the concept in this case. Space
constraints prevent further discussion of other problems with
this argument.
K&M emphasize overlap among mental illnesses, but their argument that bipolar disorder is correlated with depression is moot.
Many people with a diagnosis of major depression later have a
hypomanic or manic episode and are then correctly reclassified.
In regard to the genetic overlap of schizophrenia and psychosis
associated with mania, the question is unresolved. The phenotypic
behavior of the two states is similar, but correct differential diagnosis is crucial for the patient’s effective treatment.
When asked why bipolar disorder genes survive in the gene
pool, James Watson replied, “Survival might often depend on
not if we think two and two is four, but on being slightly
wild. . . . I think when we do science we see that a little
madness does help, and you propose bizarre things which everyone says can’t be true” (Jamison 2005, p. 275). Theory is meant to
generate new ideas, and the theory I have proposed about the
evolutionary origin of bipolar disorder generates many new
ideas. For example: Could public health measures that promote
timely exposure to more light lessen the incidence of depression
in children and adults? Does exposure to bright light during
nighttime destabilize the circadian clock and stimulate
pathophysiology among those with bipolar disorder susceptibility
alleles? Would daytime bright light and activity combined with
nighttime dark and sleep help stabilize those with bipolar
Adaptationism and medicalization: The Scylla
and Charybdis of Darwinian psychiatry
Alfonso Troisi
Department of Neuroscience, University of Rome Tor Vergata, Rome 00161,
[email protected]
Abstract: The target article shows that the application of the evolutionary
theory to psychopathology should not necessarily consist in finding
hidden adaptive benefits for each psychiatric syndrome. However, in
rejecting lax adaptationism, Darwinian psychiatrists should not forget
that the search for adaptive behavioral polymorphisms can be a
powerful antidote against the normative attitude of mainstream
psychiatry and its growing tendency to medicalize human diversity.
For evolutionary psychiatrists, the existence of psychopathology
is perplexing. How can Darwin’s theory of natural selection
explain the persistence of psychopathology that places individuals at a reproductive disadvantage? In order to resolve the
paradox of common and heritable mental disorders, most evolutionary psychiatrists have endorsed the adaptationist program.
In evolutionary biology, the adaptationist program is a research
strategy that seeks to identify adaptations and the specific selective forces that drove their evolution in past environments.
Although everyone agrees that organisms have adaptations,
adaptationism as a research strategy has not enjoyed consensual
affection within evolutionary biology (Andrews et al. 2002). In the
1970s, it became the target of attacks by paleontologist Stephen
Jay Gould and geneticist Richard Lewontin, who argued that
adaptive explanations given for most human behavioral and cognitive traits were analogous to Rudyard Kipling’s “just-so” stories.
In effect, adaptive stories are easy to create and hard to falsify,
and the risk of using inappropriate or insufficient standards of
evidence for identifying adaptations and their functions is
always present in the evolutionary study of human behavior.
Such a risk is even higher in Darwinian psychiatry, where it is
becoming more and more popular to pick a disorder out from
current nosologies (which are of dubious validity) and to invent
an adaptive explanation for its existence. Evolutionary psychiatrists have been certainly inventive in their search for a hidden
genetic advantage to mental disorders. Genes for schizophrenia
might have made certain individuals more charismatic and shamanistic. Distractible, risk-taking individuals who are currently
diagnosed as having attention deficit/hyperactivity disorder
(ADHD) might have had a competitive advantage in dangerous
environments where survival would depend on being response
ready. Anorexia nervosa might have evolved as a strategy for
conserving reproductive resources in environments in which
males are perceived as scarce and female competition for
males is perceived to be intense. The unfortunate result of
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
such a proliferation of adaptationist hypotheses is that clinicians
are skeptical of the validity and utility of Darwinian psychiatry
and believe that the preferred activities of evolutionary psychiatrists are storytelling and the invention of outlandish explanations
to elucidate the selection pressures that forged psychopathological traits (Dubrovsky 2002; McCrone 2003).
Keller & Miller (K&M) argue “against those evolutionary
thinkers who assume that adaptive forces are the only possible
explanations for common, heritable polymorphisms such as
mental disorders” (sect. 8, para. 1) and suggest that the apparent
evolutionary paradox of the existence of mental disorders can be
resolved by recognizing the enormous mutational target size of
human behaviors. K&M convincingly demonstrate that the
application of the evolutionary theory to psychopathology
should not necessarily consist in finding hidden adaptive benefits
for each psychiatric syndrome. They should be commended
because, in doing so, they avoid the risk of throwing out the
baby with the bath water. K&M acknowledge that “the search
for possible adaptive functions of mental disorder symptoms . . .
is an important counterbalance to the prevailing assumption
that subjective distress equals biological disorder” (sect. 8,
para. 2). This is a crucial point because, if adaptationism is the
Achilles’ heel of Darwinian psychiatry, medicalization is the
original sin of past and present psychiatry.
During the past two decades, psychiatric epidemiological
studies have contributed a rapidly growing body of empirical
knowledge on the prevalence data for mental disorders. Two
large community surveys conducted in the United States – the
National Institute of Mental Health Epidemiologic Catchment
Area Program (ECA) and the National Comorbidity Survey
(NCS) – showed overall 1-year mental and addictive disorder
prevalence rates approaching 30% and lifetime rates approaching
50%. This means that, according to current diagnostic criteria,
one out of every two persons will suffer from a mental disorder
during his or her lifetime. These implausibly high prevalence
rates have led to concerns about the validity of the current
methods of psychiatric diagnosis and have reinvigorated the
debate about the concept and definition of mental disorder
(Troisi & McGuire 2002). According to many, the most basic
problem with current criteria of psychiatric diagnosis is that
they fail to distinguish mental disorders from “problems in
living,” that is, the vast array of problematic but nondisordered
human conditions that reflect “the aches and pains of normal
life” (Chodoff 2002, p. 627).
In psychiatry, the medicalization of the human condition is a
long-standing problem. In the past, cultural prejudices and political aims were the major causes of the social construction of
mental illness, as, for example, in the (in)famous case of drapetomania (the “mental disease” causing black slaves to run away)
(Cohen 1981). In the last few years, the social construction of
mental illness has partly been replaced by the corporate construction of psychiatric disorder. Pharmaceutical companies are
actively involved in sponsoring the diagnostic definition of new
diseases and promoting them to both prescribers and consumers
(Moynihan et al. 2002). It seems that some new behavior is medicalized every day. The recent psychiatric literature has witnessed
the conversion of sexual desire into an “addiction” complete with
support groups, and excessive fear of social interaction into a
mental disorder treatable with medication. One study
(Laumann et al. 1999) has coined the term female sexual dysfunction to refer to sexual difficulties such as lack of desire for sex,
anxiety about sexual performance, and problems with lubrication
that are present in 43% of American women aged 18 – 59, even
though leading sex researchers have argued that these difficulties
may reflect healthy and functional responses in women faced
with stress, fatigue, or threatening patterns of behavior from
their partners (Bancroft 2002).
The necessity of avoiding medicalization of human behavior is
not simply an intellectual exercise. The concept of disease acts
not only to describe and explain, but also to enjoin to action.
For this reason, labeling a psychological or behavioral condition
as sick may have serious individual and social consequences. At
the individual level, inappropriate medicalization carries the
dangers of self-reproach, social stigma, inappropriate treatment
decisions, and iatrogenic illness. At the social level, the resources
invested in diagnosing and treating healthy people are likely to be
diverted away from preventing and treating individuals with real
diseases. Since a major contribution of evolutionary theory is the
insight that individual differences are core biological features of
any animal species, including Homo sapiens, the application of
the concept of adaptive behavioral polymorphisms to psychopathology can be a powerful antidote against the normative attitude of mainstream psychiatry and its growing tendency to
medicalize human diversity (Troisi 2005).
In Greek mythology, Scylla and Charybdis were two immortal
and irresistible monsters who beset the narrow waters of the
Strait of Messina, destroying ships as they attempted to navigate
through. The two sides of the strait were within an arrow’s range
of each other, so close that sailors attempting to avoid Charybdis
would pass too close to Scylla, and vice versa. The phrase between
Scylla and Charybdis has come to mean being in a state where
one is between two dangers and moving away from one will
cause you to be in danger from the other. The Scylla and
Charybdis of Darwinian psychiatry are adaptationism and
medicalization. To be integrated into mainstream psychiatry
without leaving behind its original contribution, Darwinian psychiatry will have to navigate through these equally misleading
Population genetical musings on suicidal
behavior as a common, harmful, heritable
mental disorder1
Martin Voracek
Department of Basic Psychological Research, School of Psychology,
University of Vienna, A-1010 Vienna, Austria.
[email protected]
Abstract: Suicidal behavior is an interesting blank space in Keller &
Miller’s (K&M’s) population genetical account on explaining the
existence and persistence of common, harmful, heritable mental
disorders. I argue that suicidal behavior is yet another of these
disorders. It may well be consistent with all three evolutionary models
considered by K&M.
First of all, I congratulate the authors, Keller & Miller (K&M),
for their population genetical analysis, concerning the existence
and persistence of common, harmful, heritable mental disorders;
it is an outstanding contribution that without doubt will have a
marked impact on the research fields addressed. One interesting
omission (or avoidance) in the target article, however, is suicidal
behavior. Only in their opening paragraph do the authors
mention that “many people with schizophrenia kill themselves”
(sect. 1, para. 1). Otherwise, the text is silent on this theme.
This commentary addresses this blank space. In order to stimulate further inquiry along these lines, I briefly discuss whether
suicidal behavior fits into the three population genetical scenarios
evaluated by K&M for mental disorders in general. For simplicity, the focus is on completed suicide.
Is suicidal behavior a “common, harmful, heritable mental
disorder,” as defined by K&M? I think the answer is yes.
Suicide is relatively common: Across many societies, it is the
tenth-or-so leading cause of death (Mann 2003), with a lifetime
mortality of around 0.5% in the general population (Bostwick &
Pankratz 2000). Suicide is harmful: It ends the suicide victim’s
physical existence, as well as any chance for future reproductive
success, and in most situations adversely impacts on the victim’s
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
relatives. Risk factors for suicide are heritable: multiple lines of
evidence – family, twin, and adoption studies (Baldessarini &
Hennen 2004; Brent & Mann 2005); molecular genetical
studies (Li & He 2006), supplemented by suggestive findings
from geographical studies (Marušič 2005; Voracek & Formann
2004; Voracek et al. 2003) – converge to this conclusion (Bondy
et al. 2006). And, finally, suicidal behavior may well be a distinct
mental disorder: It is increasingly regarded as an independent,
possibly dimensional, nosological entity (Leboyer et al. 2005).
This emerging view is based mainly on genetic evidence that its
transmission is noticeably independent of the transmission of
the risk for mental disorders. Like these, suicides also run in
families, but not necessarily in the same families; or, if so, they
may take a different path through the pedigree (Brent & Mann
2005). Whereas alcohol-related, schizophrenia-spectrum, and
mood disorders (particularly, depression) constitute major risk
factors for suicide and frequently are ascertained among suicide
cases, the vast majority (up to 80% – 90%) of people affected by
these disorders do not commit suicide (this is even valid for
suicide attempt as a risk factor for subsequent suicide).
In what follows, I argue that the current suicidological knowledge appears consistent with all three evolutionary models considered by K&M, not just with the one they opine to be the
most plausible (i.e., the mutational selection model). I commence
with this one: polygenic mutation-selection balance. According to
this model, among others, a multitude of susceptibility alleles for
suicidal behavior should be expected, all of them having small
effect sizes and many of them being population specific, thus
resulting in slow gene-hunting progress, characterized by frequent replication failures. All of this is indeed true with respect
to the search for genetic risk factors for suicide. No major
“suicide genes” have been found, polymorphisms of suspected
vulnerability genes account for small effects only and show population differences, and nonreplications are frequent in this area
(Bondy et al. 2006). The serotonergic system has been strongly
implicated in the neurobiology of both depression and suicide.
Thus, past research has almost exclusively focused on this
system’s genetic bases (i.e., on the serotonin transporter gene,
the genes for the various types of serotonin receptors, the monoamine oxidase A gene, and the two tryptophan hydroxylase
genes). Because of this theoretically and clinically well-founded
candidate-gene approach, genome scans have not been conducted in this area until quite recently, and the evidence from
a first major genome scan (Cheng et al. 2006) was disenchanting:
Several significant and suggestive linkage signals were found for
suicidal behavior in a large pedigree sample of patients with
bipolar disorder, but none of these linkage findings overlapped
with any of the gene loci of the aforementioned serotonergic
system constituents. This is expected under the polygenic
mutation-selection balance scenario of mental disorders, and in
this respect the evidence for genetic risk factors for suicide is
consistent with this model.
What about the ancestral neutrality model of mental disorders?
For several reasons, K&M do not give much credence, in general,
to this model of neutral evolution of mental disorders. Their
theoretical exception is that it could be a viable explanation for
some mental disorders which show strong genotype-environment
(G E) interaction (as indicated through large cross-cultural
variation in prevalence rates; recent historical changes in the
rates, concurring with environmental changes; and mismatch of
ancestral and modern environments). K&M mention depression
as a possible candidate for this scenario. The aforementioned
three characteristics also apply for patterns of suicide. More generally, it has been reasoned that GE interaction might be
common in psychopathology rather than rare (Moffitt et al.
2006). Suicide may be regarded as an analogue to late-onset disorders (i.e., the majority of suicides occur after the reproductive
phase). Consequently, one sociobiological theory of suicide (de
Catanzaro 1980, 1981) assumed that suicide frequently might
not be highly maladaptive. Natural selection can act only on the
residual reproductive potential. If this is low or nil, then there is
little that natural selection can affect. Further, given the large
increase in the average life span in modern environments, agerelated expressions of vulnerability genes for suicidal behavior
imply that they are now expressed under conditions different
than those they were selected for in ancestral environments.
The ancestral neutrality model might thus also have a role in
explaining the persistence of suicidal behavior.
And what about the balancing-selection model? De Catanzaro
(1980, 1981) noted the remarkable conformity of the social
ecology of suicide to the pressures of natural selection (i.e.,
relations with sex, age, cohort, and fertility, and with group,
health, and relationship status). He pointed out that many
suicides tend to occur when the anticipated residual capacity to
promote inclusive fitness is low and when staying alive could
actually reduce an individual’s inclusive fitness (particularly by
being a burden to the kinship). This perspective of suicide
appears consistent with the balancing-selection model; that is,
the view that susceptibility alleles may sometimes increase
(inclusive) fitness. Yet another variant of this model should also
not be discounted, the logic of which unfolds as follows.
Suicide is a biological anomaly: It is species general in humans,
but absent in other species. It would therefore be surprising if
certain higher cognitive traits, unique to humans, had no role
in this unique behavior. It is difficult for natural selection to eliminate deleterious X-linked alleles (for which males are hemizygous) or antagonistically coevolved genes (Rice & Holland
1997). Early attempts to demonstrate X-linkage of intelligence
(Lehrke 1997) have been met with skepticism, but it is now
clear that the X chromosome harbors an excess of genes relevant
for cognitive abilities (Zechner et al. 2001). Interestingly, two
constituents of the serotonergic system (monoamine oxidase A
receptor gene, Xp11.23; and serotonin receptor type 2C gene,
Xq24) and the androgen receptor gene (Xq11– 12) also reside
on the X chromosome. Is this mere coincidence? Or could this
imply that susceptibility alleles for suicide perhaps happened to
coevolve with positively selected variants for higher cognitive
abilities and with male-typed traits sensitive to testosterone? If
true, this would throw new light on the robust sex difference
seen in suicide prevalence (many more males than females
commit suicide). I have not yet encountered this type of conjecture in the literature, but it is testable. Nineteenth-century
suicide researchers were well aware of the link of higher
education (a proxy for intelligence) and increased suicide risk
(Morselli 1881). Using intelligence data or proxies thereof, this
pattern has been replicated in recent geographical studies,
both cross-nationally (Voracek 2004; 2005a) and intranationally
(Voracek 2005b; 2006b; 2006d), as well as in cohort studies
(Voracek 2006a; 2006c).
1. This commentary was prepared in partial fulfillment of the author’s
D.M.Sc. degree requirements at the Medical University of Vienna.
High mental disorder rates are based on
invalid measures: Questions about the
claimed ubiquity of mutation-induced
Jerome C. Wakefield
School of Social Work, New York University, New York, NY 10003.
[email protected]
Abstract: Three reservations about Keller & Miller’s (K&M’s) argument
are explored: Serious validity problems afflict epidemiological criteria
discriminating disorders from non-disorders, so high rates may be
misleading. Normal variation need not be mild disorder, contrary to a
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
possible interpretation of K&M’s article. And, rather than mutationselection balance, true disorders may result from unselected
combinations of normal variants over many loci.
Keller & Miller’s (K&M’s) impressive article will set a framework
for future discussion of the genetics of mental disorders. Their
discussion underscores the fruitfulness of understanding
mental disorders as failures of evolved functions (Wakefield
1992; 2005). I consider three reservations about their argument.
Do current rates really reflect ancestral rates of true
disorder? The first reservation concerns disorder rates. K&M’s
analysis is premised on relatively high rates of mental
disorders; the higher the rates, the more puzzling it is that
susceptibility alleles exist. However, rates may be much less
than they seem.
The rates generally come from recent psychiatric epidemiological community studies (e.g., Kessler et al. 1994). Based on
respondents’ reports of experienced symptoms, these studies
use symptom-based diagnostic criteria scored by computer to
identify disorders. While such criteria are constructed to reflect
clinically severe cases, the same symptoms may appear in individuals experiencing intense normal reactions to stress (Horwitz &
Wakefield 2006; Narrow et al. 2002; Regier et al. 1998; Wakefield
1999b). This is potentially true even in severe categories; for
example, hallucinated visits from a recently deceased spouse,
common in many cultures and individuals, may be misclassified
as schizophrenia; and reports of elation and irritability taken
as indicative of mania may in community samples represent
periods of normal elation or of irritability from marital
The fact that rates of each disorder are small compared with
non-disorder imposes brutal demands on specificity of measures
to avoid false positives. Even a small error rate in recognizing
non-disorders can yield enormous increases in apparent rates
of disorder. For example, a true rate of 0.5% and perfect sensitivity in detecting disorders combined with an unusually high
95% specificity in recognizing non-disorders still yields an
apparent rate of about 5.5%, a 1,000% increase over true prevalence. The same symptom-based methods are generally used
internationally, so international comparisons suffer from related
problems. Moreover, the threshold for a condition being
harmful may move along the symptom continuum substantially
as social circumstances change; for example, a schizotypic
individual tending to isolation may have difficulty functioning
in our society but in an Eskimo village could do well in the role
of hunter, requiring weeks at a time alone away from the
village (Wakefield 1994).
Symptom-based criteria also tend to confuse various kinds of
mismatches between normal genetic makeup and current
environments with true disorder, inflating rate estimates. For
example, as the authors note, depressive responses were likely
selected under conditions of high social support, and under
such conditions episodes may have tended to terminate rapidly;
but normal episode duration may be longer in current social
environments lacking such support and thus be mistaken for
disorders. Such mismatches could constitute a substantial percentage of depressive episodes (Horwitz & Wakefield 2005).
Similarly, what today is classified as social phobia likely encompasses naturally selected fears now seen as inconvenient, given
the mass-communicational demands of many occupations, so
that formerly adaptive aversions to, say, speaking before large
audiences, are now misclassified as disorders (Wakefield et al.
2005a, 2005b). To take a physical analogy, there may be
genetic variation bearing on the efficiency of utilization of
certain B vitamins, none of which were disadvantageous in a
natural environment in which diets were rich in vitamins, but
some of which yield vitamin-deficiency disease in vitamindepleted modern diets; the deficiency is a disorder, and it has a
heavy genetic-susceptibility component under current conditions, but if one puzzled about high rates of vitamin-deficiency
susceptibility genes and speculated about mutation-selection
balance, one would be barking up the wrong theoretical tree.
Problems with inflated disorder rates pose challenges to behavioral geneticists in general, and to K&M in particular.
Normal variation or ubiquitous disorder? A second reservation
concerns the understanding of normal variation. There is a
possible direction of thinking about normal variation as also
due to mutation-selection balance, toward which K&M might
be interpreted as moving in comments near the end of their
article. Even if they disavow this extension of their thinking,
others may interpret their argument in this way. According to
this view, we all suffer from multiple mutations that reduce our
match to a species-typical Platonic ideal to which fixation has
tended but which is not fully manifested because of many
pesky mutations. This scenario seems to potentially reduce
much normal variation to mild biological dysfunction
(Wakefield 2005) and thus to disorder, a view that may be
consistent with recent “positive psychiatry” claims that health is
an ideal that is more than mere lack of overt disorder (Vaillant
There is hopefully a more pluralist alternative. Leaving aside
the obvious possibility that trait levels confer equal fitness,
consider that excellence within normal variation may reflect the
fit between multiple components that have normal variants,
none of which are exceptional in themselves. For example,
Eclipse, a chestnut colt foaled during a total solar eclipse in
1764, is generally considered the greatest racehorse of all time.
Autopsy results showed a large heart and lungs, but what leg
structure enabled Eclipse to run so exceptionally fast? The
answer, according to a recent computer-simulation analysis by
members of the Structure and Motion Laboratory at the Royal
Veterinary College based on Eclipse’s preserved skeleton, is
that Eclipse’s various leg bones were all remarkably average
and thus fit together in an optimally cohesive package (R.
Weller, personal communication). Horses that are slower are
not suffering from mutations but from a less striking confluence
of normal variants over a large number of loci. Normal variation
of mental traits may work like that: normal variants at multiple
loci on which the trait is based (here we must accept the K&M
“watershed” model) yield phenotype variation because of
routine assortment.
Disorder as nonselected combinations of selected traits.
Approaching normal variation from the above perspective
suggests an analogous alternative to Keller and Miller’s
mutation-selection balance explanation for presumed high
mental-disorder rates. For example, rather than the antisocial
personality being frequency selected to fill an adaptive
“cheater” niche or caused by accumulating mutational
disadvantages at many loci, levels of conscientiousness, thrill
seeking, and other personality dimensions may happen to
converge in some individuals in a toxic combination of
components, each of which is a normal variant. Together,
these variants may add up to a phenotype that was not part
of the range that yielded selection of the involved normal
variants (Wakefield 1999a). Much pathology could be such
normal-variant combinations falling outside the selected zone
of multiple-loci combinations. Such a model is consistent with
findings such as those of Eaves et al. (1990) that neither
neuroticism nor extraversion alone had fitness implications,
whereas combinations of specific levels of each did have
differential fitness results.
Is this just a resurrection of the “non-additivity” alternative
that K&M reject? To some extent, but their comments on nonadditivity seem to me the most thinly supported within their
argumentational juggernaut, leaving this perhaps a viable direction for exploration. Moreover, some balancing selection or
other constraints may be at work at various levels of the
genetic “watershed,” not directly selecting for the disordered
combination of traits but rather exerting selective force to keep
the variations on individual traits (e.g., low conscientiousness,
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
high thrill seeking) in the population. Consideration of which
such scenarios yield susceptibility alleles “invisible” to or ineliminable by natural selection seems a promising direction in
which to look for viable alternatives to K&M’s mutation-selection
balance thesis for some disorders.
Multiple timescales of evolution
Jonathan Williams
Department of Child and Adolescent Psychiatry, Institute of Psychiatry,
London SE5 8AF, United Kingdom.
[email protected]
Abstract: Keller & Miller’s (K&M’s) treatment of disorders usefully
avoids diagnostic minutiae; but it needs more real-world constraints.
Classifying processes by their evolutionary age helps to clarify both
evolution and current function. Evolutionarily old, optimised,
normative processes deserve special recognition, because they can be
studied in animals and computers, and because they provide the
machinery through which disorder-related polymorphisms act.
Introduction: Dimensions of the task. Keller & Miller (K&M)
see evolution as a continuous series of mutations, competing for
selection in a distant, rather abstract world. Other scholars focus
on brain developments that were taking place at the moment in
prehistory when a disorder became recognisable (e.g., Crow
1995b). Yet others reduce evolution to a “final cause” or
purpose for each disorder (e.g., Jensen et al. 1997; McGuire
et al. 1992).
These authors each address just one aspect of causation, and
neglect the whole (see Killeen & Nash 2003; Pearl 2000). The
causal networks underlying the appearance of a disorder in an
individual at a particular point in time are more complex than
any river (contrast K&M’s Fig. 3 with Kendler et al. 2002,
Ramus 2004, and Sonuga-Barke 2005).
Crucially, causes of normal function must be included within
these causal networks, not in the “mutual” relationship K&M
suggest. Addressing mental disorder without normal function is
like practicing medicine without physiology (Nesse 1991; cf.
Dayan & Williams 2006). Not only are many mental disorders
“expressed through the structural medium of normal brain function” (David & Halligan 2000, p. 509), many are distortions of
normal function. Normal functions such as learning (Moore
2004), language (Elman 2005), and walking upright (Gregory
1928) – as well as, doubtless, emotion and cognition – do not
arrive along a tidy trajectory: they arise and change through
many stages, and persist or fail for many interacting reasons;
and many abnormal functions must be as complicated. K&M
cut through this complexity, offering the long-accepted principles
of polygenicity, mutation, and selection as the basis for all
common mental disorders; but differences between disorders
are the real issue, and linking evolution to the remarkable
recent progress in psychiatric genetics requires considering
genes not as abstractions but as having specific neural roles in
bodies interacting with real environments.
K&M mention optimization in their first sentence, but do not
consider its ramifications. A behavioural response that is optimal
in one environment may be disastrous in another; and occasionally organisms need to make mistakes in order to see whether the
optimum has changed (Williams & Taylor 2006). Such mistakes
are a form of entropy, which K&M distrust, stating that
“entropy erodes functional complexity” (sect. 1.3, para. 3), but
entropy is also a prerequisite for genetic and social evolution
(Williams 2005). Regulation of this entropy, a meta-optimisation,
is seen in lower animals, which have sophisticated mechanisms
“specifically to increase the rate of mutation in localized parts
of the genome” (Metzgar & Wills 2000, p. 584) – as do humans
(Chuang & Li 2004). Such careful regulation hardly merits the
term “inevitable mutational load.”
An organising heuristic. Some of these problems can be
avoided by classifying brain processes by their approximate
time of evolutionary appearance. Putting measurements into
broad time bins is a standard heuristic for disentangling
complex biological causation (Samoilov et al. 2001; Wen et al.
1998). Boxes in Figure 1 contain causes characterized by
duration, but also correlated with prevalence, manipulability,
study methods, and the level to which they have been
subjected to mutation and selection (optimisation). The
stacking of boxes indicates the accumulation of influences from
bottom to top, with each layer grounded, like sediment, on the
layer below it. Hence, mutations unique to an individual (f)
generally exert their influence via long-evolved systems of
which they are part (b and d).
The suggested categories are:
a. Universal experiences, available even to primitive animals.
These include the clustering of signals in the environment, the
reduction in signals from distant sources, gravity, and diurnal
rhythm (Campbell 1974).
b. Normative neural processes (which K&M call “universal
features of human nature”; sect. 5.1, para. 1) acquired due to category (a), such as primitive learning and regulation of exploration. Pragmatically, these deserve to be distinguished from
K&M’s abstract “inevitable mutational load” because they are
highly conserved (Campbell 1974; McAdams & Pals 2006;
Moore 2004) so can be studied in animals and can be presumed
present in humans even if submerged or distorted; also they can
be presumed to be largely optimised so predictions from
optimised mathematical processes can be used to reverse engineer them (e.g., see Gadanho 2003; Montague et al. 1996).
c. Species-specific experiences, and common differences in
experience, such as between hot and cold climates or wealth
and poverty, or experience of social parasitism (discussed by
K&M in sects. 5.6 and 5.8).
Figure 1 (Williams). Some broad categories of influences on an
individual’s current cognition and behaviour. Factors common to
all mammals can be included in layers a & b. Additional layers
could be added within layers c & d for nested groups such as
primates or human subgroups. Adaptations within a lifetime
have important effects on disorders (Sonuga-Barke 2005) and
even on the genome (Hinton 1987), as do consecutive
adaptations in an individual species (Richerson & Boyd 2004);
these are excluded for simplicity.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
d. Species-specific processes (e.g., language) and common
intra-species genetic variations. Several factors control these,
including non-harmfulness (discussed by K&M sects. 3.3, 4.3),
environmental differences (c), and benefits of population diversity
(Williams & Taylor 2006). Optimality is generally impossible to
ascertain here, unlike (b).
e. An individual’s experiences, both shared and unshared.
These are biological, psychological, and social; K&M mention
illicit drugs in section 4.4. As another example of the complexity
of causation of disorder, anti – basal ganglia antibodies induced
by Streptococcus are found in 50% of children in some clinics
(Dale & Heyman 2002).
f. Each human contains about 175 new mutations
(Nachman & Crowell 2000). Most of these are obviously not
significant clinically, but the proportion of disorder attributable
to them is a major unknown.
Normative processes (b) are a crucial tool in our teasing apart
disorders into conceptually coherent components. Some
instances of a disorder are fully appropriate to recent life
events (i.e. determined by normative processes: eþb in Fig. 1).
Others are the inevitable outcome of rare, functionally obvious,
mutations, which must nonetheless act through normative
processes (fþb). The majority are more complicated, and many
of these appear to result from common polymorphisms superimposed on normative processes (eþdþb), summarised in the
following paragraphs.
Depressive disorders. Conserving energy at non-propitious
times is useful to animals (a&b), and to robots during
exploration (Gadanho 2003). Such studies clarify the
evolutionary origins of mood, clearly relevant to the evolution
of mood disorder (see also Keller & Nesse 2005). So learned
helplessness is a handle for studying the neuroscience of such
normative responses that would produce much less
interpretable results if applied to abnormal animals, or indeed
to humans diagnosed with depression. Criticisms of such
models as being unlike human disorder are factually correct
but almost irrelevant to this process-based approach.
Adverse life events certainly contribute to depression, but the
31% of the population homozygous for long 5HTT-LPR appear
to be protected from this effect (Caspi et al. 2003). This and
the social importance of depression (e.g., its possible role in
honest signalling; Watson & Andrews 2002) suggest modification
in humans of the much longer-term mechanism (b).
Depression and anxiety are largely due to the same genetic
factors (b&d), yet depression tends to follow loss events (e),
whereas anxiety follows threat (Eley & Stevenson 2000;
Kendler et al. 1987). This shows a limitation of K&M’s confessed
environment-light approach. Furthermore, a full account of
depression clearly needs to describe why some people are genetically predisposed to encounter more adverse events than others.
Attention-deficit/hyperactivity disorder (ADHD). Punishment,
threat, and reward regulate activity and attention in all
organisms, as alluded to previously. Other control systems
regulate our exposure to new information, via alertness,
boredom, and attentional control (a&b). It appears that many
of the deficits associated with ADHD can be attributed to
aberrations of these normative control systems (Williams &
Taylor 2004) or of the similarly ancient neural mechanisms of
decision-making (Williams & Dayan 2005).
On the surface, ADHD is a family of genetically determined
deficiencies with the fractions attributable to particular alleles
(mainly dopaminergic to date) ranging from 8% to 20% (Daly
et al. 1999; Faraone et al. 2001, 2005). However, there is evidence that one of its susceptibility genes is positively selected
for (Ding et al. 2002) and it now seems likely that, despite disadvantages for the individual, ADHD conveys subtle benefits for
the group (Williams & Taylor 2006). These include genetic
exploration which is closely related to K&M’s thesis, and group
risk minimisation during behavioural exploration, which is not.
Schizophrenia. Despite this disorder being relatively rare,
disabling, and characteristically human, its causation clearly
involves long-evolved processes (b) including synaptic pruning,
fear, emotion regulation, sensory filtering, causal attribution,
and the increased focussing of attention when anxious
(Cosmides & Tooby 1987; Luce et al. 1997).
Schizophrenia also involves recently evolved human characteristics (d) such as cerebral lateralisation and language (Crow
1995b). Similarly, cognitive dissonance probably contributes to
delusional mood, delusions, and hallucinations (Brabban &
Turkington 2002), but is less important in nonhumans (Armus
The developmental instability hypothesis (Yeo et al. 1999)
suggests that schizophrenia and other neurodevelopmental
disorders result from an individual’s reduced ability to “buffer”
developmental insults. The deficiency preferentially affects
males, who have four times the nucleotide mutation rate
(Nachman & Crowell 2000) and a similar excess of neurodevelopmental disorders. This sex difference is a key aspect of the evolution of mental disorder (see Williams & Taylor 2006) without
which a model as abstract as K&M’s appears impoverished.
Conclusion. Adding to previous examples, this quantitatively
demonstrates that the multidimensional space of genetic
possibilities is explored not indiscriminately, but by branching
“tentacles” which incrementally, tentatively, and interactively
reach up through the levels shown in Figure 1. In this
metaphor, the two hypotheses dismissed by K&M probably
describe some tentacular branchings. Their favoured hypothesis
has few constraints beyond Watson and Crick’s, and
successfully places the octopus in a large bag. Now we need to
get inside and study the octopus’s normative processes.
I thank Peter Dayan, Peter Killeen, Michael Moutoussis, and Eric Taylor
for helpful comments on a draft of this commentary.
The evolution of evolutionary epidemiology:
A defense of pluralistic epigenetic modes of
Daniel R. Wilson
Departments of Psychiatry and Anthropology, Creighton University, Omaha,
NE 68131.
[email protected]
Abstract: First kudos, followed by some friendly badinage, and then
renewed appreciation and a look ahead. This commentary is meant to
clarify main arguments, redress incorrect attributions, and strengthen
an excellent contribution that draws further attention to the importance
of evolutionary epidemiology. Keller & Miller (K&M), despite
significant errors, have done well to further systematize the
evolutionary epidemiology of psychopathology.
The target article, which helps to reform evolutionary psychology, begins lucidly as Keller & Miller (K&M) re-emphasize
three basic ways in which selection can operate on etiogenes
associated with common, harmful, heritable mental disorders
(MDs), for example, schizophrenia, bipolar disorder, phobias
and mental retardation. The article later elegantly extends into
a more robust evolutionary epidemiological framework, but
then falters amid the murky mires of psychiatric genetics.
Still, the review is an impressive scholarly synthesis of many
pertinent concepts and relevant information that, taken together,
systematizes how to assess evolutionary epidemiology of major
mental illnesses. It is a way forward as sufficient, necessary,
and naturalistically valid data may accrue to really understand
each epigenetic syndrome. But not only has nothing so much
yet accrued, there are other problems.
Commentary/Keller & Miller: Resolving the paradox of heritable mental disorders
Darwin saw valid taxonomy as absolutely essential to evolutionary cogency. Therefore, by way of first and strongest
criticism, it must be said plainly that the main fulcrum of
argument, “mental disorder,” is a fallacious, intrinsically nonLinnaean pseudotaxon. This construct so broadly conflates
distinct moieties – schizophrenia, depression, mania, diverse
phobias, and even mental retardation – that it is ultimately
without demonstrable naturalistic validity.
I, with others, have long held that the received diagnostic
categories are often incompatible with genetic research, and I,
with others, have urged more parsimonious nosology with
better lumping of phenotypic equivalents. But this proffered
straw man, “mental disorder,” is – literally – chimerical and
inadequate to the task. From this anheuristic notion, one
simply cannot cobble together the diverse building blocks for
any deep understanding of psychopathology in its wide remit.
For example, most mental retardation is caused by chromosomal
anomalies that have, strictly speaking, nothing to do with genetic
Similarly, the authors often over-argue to a preferred mechanism and, in so doing, misconstrue others’ prior work. For
instance, John Tooby and Leda Cosmides, as well as Linda
Mealey, are incorrectly attributed to have believed that genetic
variation in human psychology is the result only of neutral or
balanced selection. Rather than the clumsy sweep wrongly
ascribed to them, these researchers have noted this only with
specific reference to a few traits of interest. No one (with the
possible exception of Tim Crow and his “psychosis gene”; cf.
Crow 1995a) has recently argued for a Mendelian single-locus
model of schizophrenia, bipolar disorder, phobias, mental retardation, or much else.
My own work is likewise battered about on several occasions,
and usually badly. For example, early on in the target article I,
along with other psychiatrists interested in evolutionary explanations, am given the backhanded compliment that “clinicians
more familiar with psychiatric hospitals, prisons, and detox
centers were understandably skeptical . . . [of] Panglossian evolutionary ideas” of Darwinian psychiatrists (sect. 2, para. 2).
This would-be ad hominem is most amusing! I have spent my
entire career in extremely active hospital care of persons suffering from psychoses, mania, depression, substance abuse, and
nearly all else. Indeed, I have worked at the sharpest edges of
hospital, public, and forensic psychiatry, having admitted in
excess of 4,000 acute manics or psychotics at Harvard affiliated
McLean Hospital (Belmont, MA) and later committed some
3,000 dangerous persons as medical director of Summit Behavioral Health Center in Cincinnati, the largest (and most poorly
funded!) state hospital in Ohio.
So, too, Nesse, Price, McGuire, Gardner, Gilbert, Beahrs,
Sherman, Pearce, Erickson, Sloman, Thompson, and others of
Darwinian stripe are not naifish armchair psychiatrists befogged
in Ivory Towers. Moreover, I, like most psychiatrists of Darwinian leaning, do not misunderstand selection-mutation nor
espouse Panglossian adaptationism nor wallow in miscreance,
as the authors so casually indict! Ah well, a side issue perhaps.
But no, as then again K&M totally invert my long-standing and
oft-stated view that a great deal of, and perhaps most, psychopathology is the stochastic noise of mutation selection. Still
later (sect. 5, para. 1), I am said to have implied that balanced
selection is the only plausible explanation for prevalence of
genes linked to common psychiatric disorders. I have only said
that of highly prevalent quasi-Mendelian traits and meanwhile
have always held that neutral evolution and mutation-selection
balance presumably explain many other syndromes. Thereafter
again, analogy is taken literally, as I am attributed to have stipulated that heterozygote advantage à la sickle cell is the sole mechanism that sustains epigenes for bipolar disorder (see sect. 5.4,
para. 2). Oh dear!
Meanwhile, game theoretical analysis of clinical phenomenology is an important and valid corpus of evidence that the authors
either dismiss or ignore, even though it is germane to resolving
the possible paradox of common, harmful, heritable mental disorders. Perhaps it constitutes some of the “tortuous . . . frustratingly implausible” stuff referred to early on (sect. 1.1, para. 3).
But game theoretical mathematics and the neuroethology of
social rank and mood are not only the stuff of Nobel Prizes,
they are also heuristics well beyond the capacity of the “mental
disorder” construct to address with any fidelity or insight
(Gardner & Price 1999). Oh dear, oh dear!
These broad-brush tarrings too often detract from K&M’s
otherwise elegant article. Though I cannot speak for others,
these are strong misreadings of, if not liberties taken, with the
main body of my work. This work has for some 20 years
focused on the extended phenotypies of bipolar disorder as one
of the rare psychiatric diagnoses that (1) approaches validity as
a natural taxonomic unit; (2) plausibly transmits an adaptive
endophenotype consistent with oligenic, kinship, and negative
frequency-dependent selection; (3) clearly corresponds to
social rank and stress biology phenotypies common to social
mammals, including the limbic system and stress axis, especially
in the anthropoid line; (4) is testable as specific game mathematical cost/benefit consequences of neuroethological variants in
reptilian, paleomammalian, and neomammalian brain and behavioral repertoires; (5) entails a fascinating range of corollary traits
of relevance to sexual selection and assortative mating, among
many other ethological and sociobiological factors; and (6) may
express more pathophenotypy amid the rampant mismatches
and disruptive genomic reactivity of modern developmental
The authors have done nothing to undermine this line of
Yet all that is beyond the present scope, so then back to the
other major flaws in the article beyond the untenable construct
of “mental disorders.” The further claim is made that these are
mostly extreme points along axes of dimensional expression.
This suggestion repeats a current fad among genomic researchers
who, unable to easily replicate preliminary linkage studies, avow
this failure can only mean traits in question are necessarily
complex, polygenic, or even Gaussian in their distribution, and
thereby without basis, presume these can only be extreme tips
of normative dimensional traits.
Alternative interpretations are plausible and more accurate,
although in the present state of genetic knowledge it must be
plainly said that we simply do not know. Key issues such as
mutation, fertility, homology, mismatch, and heterogeneity are
almost entirely inferential. Indeed, the classical genetic studies
of twins, adoptees, and families still inform more about several
major psychiatric syndromes than does molecular work and
these have long suggested prevalence about likely mutation
rates (Wilson 1998).
Despite this absence of critical data there is also generally, and
in the target article, an unfortunate tendency to terminologically
assume what is not Mendelian is necessarily complex and dimensional. What of oligogenesis? What of endophenotypy? What of
threshold trait expression? What of epistasis? What of the
effects of major or even moderate genes admixed with a
varying host of minor traits? What of the vast and uncharted
realms of proteomic regulation? What of so much else?
Meanwhile, negative frequency-dependent selection is given
short shrift even as it may sustain a number of human capacities,
including several that are, in terms of possible genomic and
sociocultural dynamics, akin to the bipolar extended phenotype,
notably sociopathy (Mealey 1995), left-handedness (Brooks et al.
2004; Faurie & Raymond 2005), Tourette’s (Pauls 2003), and
homosexuality (Rahman 2005).
Negative frequency-dependent selection occurs in special situations such as in plant self-recognition, host– parasite relations,
and mimetic species – in which a tasty butterfly has two alleles,
each of which mimics a different poisonous butterfly. Here,
rarity increases fitness. So, too, negative frequency-dependent
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
selection occurs in social competition (viz., what behavioral ecologists call pairwise contest and comparative ethologists call ritual
agonistic encounter, that, by whatever name) clearly corresponds
to clinical ego states of dominant or submissive mood, affect, and
behavior. Choice of protagonist strategy depends on the strategy
opted by the antagonist rival, and, in general, if there are two
possible strategies, it pays to adopt the strategy not chosen by
the rival. Handedness is a manifestation of this in the physical
sphere – when fighting a right-hander, it can pay to be a
As in the physical domain, so in the behavioral. The two main
strategies are escalation (hawk) and de-escalation (dove). If the
rival escalates, it may pay to de-escalate. Hence, agonistic
social ranking is likely a key normative aspect of the endophenotype for bipolar disorder that is quite plausibly inherited as a
negative frequency-dependent polymorphism (whatever the
number of alleles!). The tendency to too readily over-escalate
or de-escalate rises as novel developmental environs are encountered (i.e., mismatch). Yet the target article ignores such game
mathematic analysis of frequency-dependent selection clinical
phenomenology as well as its evolved social neurobiology
Perhaps most fundamentally unpersuasive is how the target
article posits that there is (much less, that there must be) only
one mechanism of genomic transmission for a wide variety of psychiatric syndromes – wrongly subsumed into a spurious category
of “mental disorder.” A glaringly obvious question, with an
equally glaringly obvious answer, is: What do the phyloepigenetics of apples (e.g., mental retardation) have to do with
oranges (e.g., mania)? This slap-dash Linnaeanism is a taxonomic
house of cards.
Who presently can say that the clinical range of phobias
derives from a single mechanism, much less that, depression,
psychosis, or Tourette’s – and all the rest in the panoply of psychopathology – necessarily spring from a single selective mode?
As it is, improper taxonomics, notably poorly characterized endophenotypes, renders nearly all of psychiatric molecular genomics
of little present help in understanding the epigenetics of major
syndromes of psychopathology, much less their phylogeny.
It is entirely possible (and, as I have argued previously, likely)
that a range of epigenetic mechanisms express diverse syndromes. It is also entirely possible (and, as I have also argued previously, most likely) that several major syndromes are composites
of clinically homologous but genetically distinct variants. If there
are variant strains of depressions or manias or psychoses, a few
may well be Mendelian, others oligogenetic, still others major
traits epistatically affected by a range of minor interacting
genes, and yet others merely extremes of normal dimensions.
Any such variation may explain much in the way of co-morbidity.
It also would, and, as I have further mentioned previously, most
surely continues to, play havoc with molecular linkage and
related analyses.
The paradox – if it is paradoxical – of common, harmful, heritable mental disorders is not resolved. K&M’s belief that nearly
all psychopathology has a single mode of phylogeny – simple
stochastic noise – is but one possible resolution. Does a whiff
of over-enthusiastic, prematurely conclusory triumphalism
detract? Time will tell. Meanwhile, despite a number of substantive limitations, the target article has many useful implications
for psychology, psychiatry, and genetics, among other means to
study human behavior. It is especially welcome for more
clearly conceptualizing how future progress can be made in evolutionary epidemiological analyses of psychopathology.
I owe thanks to a great many colleagues and patients for helping me
elaborate these ideas, and, especially, to the late Dr. Ernst Mayr, as
well as to Professors E. O. Wilson, Nick Mascie-Taylor, John Price,
Paul Gilbert, Randy Nesse, Hagop Akiskal, Kay Jamison, Russ Gardner,
Jaak Panksepp, and Gerald Cory.
Authors’ Response
An evolutionary framework for mental
disorders: Integrating adaptationist and
evolutionary genetic models
Matthew C. Kellera and Geoffrey Millerb
Virginia Institute for Psychiatric and Behavioral Genetics, Virginia
Commonweatlh University, Richmond, VA 23219; bDepartment of Psychology,
University of New Mexico, Albuquerque, NM 87131-1161.
[email protected]
[email protected]
Abstract: This response (a) integrates non-equilibrium
evolutionary genetic models, such as coevolutionary arms-races
and recent selective sweeps, into a framework for
understanding common, harmful, heritable mental disorders;
(b) discusses the forms of ancestral neutrality or balancing
selection that may explain some portion of mental disorder risk;
and (c) emphasizes that normally functioning psychological
adaptations work against a backdrop of mutational and
environmental noise.
R1. Introduction
We thank the authors of the 23 commentaries for their
generous comments, valuable insights, and helpful suggestions. We are especially grateful to those contributors who expanded our framework in promising new
directions. As a parallel to our target article, we
organize our response to the comments according to
the different types of evolutionary genetic explanations
that can help make clear the central paradox of our
target article: What explains the evolutionary persistence of susceptibility alleles that increase the risk of
common, harmful mental disorders? The commentaries
can be separated into those that suggested processes
that did not receive enough attention in the target
article (sect. R2), those that advocated neutral evolution
or balancing selection mechanisms as viable alternatives
to a mutation-selection model (sects. R3 and R4), those
that called into question the evidence we marshaled in
favor of mutation-selection explaining some portion of
the mental disorder risk (sect. R5), and, finally, those
that made specific new points or that pushed our thinking in new directions (sect. R6). As an antidote to our
“whiff of over-enthusiastic, prematurely conclusory
triumphalism” (Wilson), we also emphasize our
omissions, errors, and confusions, where appropriate.
By necessity, some good points made by individual
commentators – especially those that we agreed with
and therefore felt less impelled to respond to – must
go unaddressed.
R1.1. The goal of our response
Our target article had two main goals. First, we wanted
to promote more consilience among behavioral/psychiatric geneticists, evolutionary psychologists/Darwinian
psychiatrists, and evolutionary geneticists. Second, we
wanted to review the models that can explain the persistence of susceptibility alleles, understand the predictions
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
those models make, and review relevant empirical evidence. In writing the target article, we came to realize
that the best-supported model, both theoretically and
empirically, was a balance between individually rare,
harmful mutations arising at the thousands of loci that
must underlie complex human behaviors, and their
eventual removal, due to natural selection. Although
several commentators called into question the evidence
supporting a role of mutation-selection in the genetic
risk of mental disorders, we did not find these arguments
particularly compelling or well supported by the data
(sect. R5).
If mutation-selection explains some portion of susceptibility alleles for a given disorder, then this has a couple of
important consequences. First, it means that no other
single explanation (viruses, sexual conflict, heterozygote
advantage, etc.) can be the sole explanation for that
mental disorder. When we say in our response that some
process is unlikely to be “a general resolution to the
paradox,” this is all we mean: that it appears insufficient
for explaining fully the existence of a disorder’s susceptibility alleles, not that the process is unlikely to explain a
portion of that risk. Contra several commentators, our
point has never been to argue that a single mechanism
explained all mental disorders (sects. 4.4, 5.8, 7.6, 8, and
R5.2). Second, as we argue in this response, we think
that a mutation-selection explanation is mutually incompatible with only two evolutionary models of mental
disorders: (a) that mental disorders are affected by only
a few genes, or that they are byproducts of “genes that
made us human” that have already fixated in the population (sect. R2.2), and (b) that mental disorders are
themselves complex adaptations (sect. R4.1). Other than
these two models, evidence for a mutational role in
mental disorder risks is perfectly compatible with other
types of evolutionary explanations, and with a general
adaptationist perspective (R5.4).
A major point in our response is to argue for theoretical pluralism. This does not mean that we have to
believe that each hypothesis has equal explanatory
weight; the evidence should be allowed to clarify their
relative importance. As we did in our target article, in
our reply we try to generate discriminating empirical
predictions for each type of process, and, when possible, we bring whatever data that we are aware of to
bear on the question. Ultimately, the validity of these
models is going to be quantitative, not qualitative. So,
for example, 35% of the risk of some given mental
disorder may stem from deleterious mutations, 20%
from environmental insults, 15% from alleles in hostparasite conflict, 10% as a side-effect of balancing
selection on other traits, and 8% as a side-effect of
alleles sweeping to fixation or extinction (ignoring
obvious complicating factors that certainly occur, such
as interactions between these processes). For other
disorders, such as those that have greatly increased
in incidence since ancestral conditions (R3.2), the mix
might be much different. Nevertheless, as things
stand, we believe that current empirical evidence best
supports a mutation-selection balance model for the
types of serious, common mental disorders that we
focused on in our target article (sect. 1.3). Time will
tell whether the weight of current evidence reflects
the true etiological mix.
R2. Evolutionary genetic processes that did not
receive enough attention in the target article
In attempting to resolve the paradox of common, harmful,
heritable mental disorders, our target article focused on
neutral evolution, balancing selection, and mutation-selection balance, which are the three principal models used by
evolutionary geneticists to explain genetic variation in
nature. These models are useful for understanding gene
frequencies that are roughly at equilibrium (stable and
unchanging) in the population. Gangestad & Yeo,
Crespi, and McGrath suggest two non-equilibrium
models that may also be relevant to understanding
genetic variation in mental disorder risk: antagonistic
coevolutionary arms-races and recent selective sweeps.
We agree that both models deserve more thorough discussion. As pointed out by Gangestad & Yeo, equilibrium and
non-equilibrium evolutionary genetic processes are in no
way mutually exclusive – both classes of processes must
occur simultaneously. In this section, we consider the
two non-equilibrium evolutionary genetic models in
turn, although, for clarity of presentation, we distinguish
two classes of coevolutionary arms-races: those that
occur between hosts and parasites (R2.1.1) and those
that occur between genes within the same species
R2.1. Antagonistic coevolutionary arms-races
R2.1.1. Antagonistic coevolutionary arms-races between
hosts and parasites. Gangestad & Yeo and Crespi
discuss situations where the human genome is not in
evolutionary equilibrium because of ongoing coevolutionary arms-races between hosts and parasites. We also
mentioned this mechanism as a possible candidate for
explaining some mental disorder risk (sect. 5.8), although
we are somewhat perplexed that we did not discuss its
relevance in more detail, given our previous work on it
(Cliff & Miller 2006; Miller 1997).
Although host– parasite coevolution can be powerful at
maintaining genetic variation at loci concerned with antipathogen defense (Hull et al. 2001; Tooby 1982), such as
the major histocompatibility complex (MHC) loci, we
argued in the target article that it is unlikely to cause
much maladaptive variation in psychological traits as a
side-effect (sect. 5.8). This is because only a small minority
of loci are directly involved in pathogen defense (Venter
et al. 2001), and because natural selection should favor
minimum overlap between the highly variable defense
system loci (in which mutation and crossover rates may
be adaptively higher; Metzgar & Wills 2000; Nachman
2002) and the tightly regulated loci that are involved
more centrally in neurodevelopment. Indeed, as one
might expect, most disorders that stem from defective
genes in the defense system are related to failures of
pathogen defenses – anomalies in immune response,
pathogen susceptibilities, failures to distinguish between
self and foreign antigens, allergies, and so forth (F.
Vogel & Motulsky 1997) – rather than to neurological or
behavioral disorders.
However, host– parasite coevolution might maintain
behavioral genetic variation in more subtle, and
interesting, ways than we discussed. How this might
occur requires some explanation. The starting place is
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
appreciating that viruses, protozoa, bacteria, and other
parasites can cause long-term dysfunction that might
otherwise appear as congenital or developmental disorders
(Cochran et al. 2000). The classic medical example
concerns stomach ulcers, which used to be classified as
maladaptive stress responses, but which turned out to be
caused by infection by Helicobacter pylori (Marshall
et al. 1988). Some human mental disorders may be
similar, as in the hypotheses that the protozoan Toxoplasma gondii (Ledgerwood et al. 2003) and in utero
viral infections (Brown & Susser 2002) increase schizophrenia risk. As noted by McGrath and Nettle, this
evidence is perfectly consistent with a model whereby
neurodevelopmental disruptions from many different
sources, including mutations, cause schizophrenia (see
also Gangestad & Yeo 1997).
The dysfunction wreaked by parasites may not simply
be an unfortunate by-product; parasites might actively
manipulate hosts to produce apparently bizarre behavior
for their own benefit. For example, rabies makes dogs
enraged so they bite more often, which spreads the
rabies virus through saliva; the rage is harmful to the
dog but adaptive to the virus (Klein 2003). Similarly,
sexually transmitted parasites in humans might provoke
promiscuity (Cochran et al. 2000), and contact-transmitted
parasites might provoke not only sneezing and coughing,
but also behavioral changes such as poor hygiene or indiscriminate social contact. If one is blind to these conflicts
between hosts and parasites, one can easily mischaracterize a parasite’s adaptation as a host’s maladaptation.
These specific ideas may be right or wrong, but parasitemanipulation explanations deserve more attention in
Darwinian psychiatry.
Are such host– parasite coevolutionary models relevant
to the central paradox of our target article? As already
noted, parasites appear to have a modest effect on the
risk of schizophrenia (and perhaps other mental disorders), but this risk could come from two different
kinds of processes. First, if population differences in the
risk of initial infection and/or subsequent suffering from
it are not highly heritable, then parasite infections would
be additive factors that increase schizophrenia risk,
pushing already at-risk (e.g., high mutation load) individuals over the edge. Some people might describe this as a
gene-by-environment (G –E) interaction, but it really is
not, because the environmental agent (the parasite)
merely adds risk to whatever genetic and environmental
risk an individual already has. This mechanism is not a
resolution to the paradox because an individual’s genetic
risk (e.g., mutation load) still requires an explanation,
which could come from any of the evolutionary genetic
models we have reviewed.
The second process, however, can provide an alternative explanation for the persistence of susceptibility
alleles. If differences in the risk of being infected and/or
suffering from infections are highly heritable, then parasite infections and susceptibility alleles would be interactive factors increasing schizophrenia risk. In this case,
individuals unlucky enough both to be infected and to
have the predisposing alleles would be at higher risk of
developing schizophrenia. Individuals carrying the predisposing alleles would also pass these “susceptibility alleles”
on to their offspring. This is a plausible G– E interaction
scenario: Susceptibility alleles are maladaptive only
(or mostly) in environments where the parasites are
endemic. This process can also explain the high heritability
of mental disorders, because alleles that are adaptive at
one time point become maladaptive later, as the parasite
evolves new ways to circumvent old defenses. Under
this scenario, susceptibility alleles can be at high frequencies, similar to alleles governed by balancing selection
(indeed, host –parasite coevolution is often considered a
type of balancing selection).
What types of evidence would support this latter
possibility as an important explanation for the evolutionary
persistence of schizophrenia susceptibility alleles? First,
this type of process predicts not only interaction effects
on schizophrenia risk, but also main effects both for parasite infections and for alleles that cause susceptibility to
those infections. While the infection-schizophrenia link
is quite strong (Sullivan 2005), there is little evidence for
genes of major effect in schizophrenia, as described in
the target article (sects. 7.6 and R2.3). This is not necessarily damning evidence because it might indicate that such
G– E interactions are only partial explanations, but it
weighs more strongly in favor of the additive than the
interactive hypothesis. Second, we should also expect
that likely schizophrenia susceptibility alleles will tend to
cluster in genome regions associated with pathogen
defense, but we are aware of no reliable supporting
evidence (Sullivan 2005; Wright et al. 2001). Thus, for
the moment, we find the hypothesis that parasite infection
is one of many additive risk factors in schizophrenia most
compelling, while acknowledging that interactions
between parasites and host’s genes could also play some
as yet to be determined role.
R2.1.2. Antagonistic coevolutionary arms-races between
genes within the same species. Coevolutionary arms-
races can occur anytime genes have conflicting goals,
both between and within species. Gangestad & Yeo
and Crespi argue that antagonistic coevolution between
genes within the same species might help explain some
portion of mental disorder risk. Although it can take
several forms (between siblings, between sexes, etc.),
here we focus on antagonistic coevolution between
maternal and paternal genes within an offspring
(Gangestad 2003; Rice & Holland 1997). The optimal
amount of maternal investment from the perspective of
paternal genes is higher than the optimal amount from
the perspective of maternal genes. This is because each
maternal gene is guaranteed to have a 50% chance of
being represented in the mother’s future offspring,
whereas paternal genes have less than a 50% chance of
being represented in her future offspring (because she
might mate with a different male).
The fact that the effects of genes can depend on
whether they were inherited from the mother or the
father, called genetic imprinting, sets up the means by
which differences in optimal strategies between maternal
and paternal genes can turn into coevolutionary armsraces. For example, paternally expressed genes might
cause infants to take a bit more milk than they did
before, which would provoke a counter-response in maternally expressed genes, causing infants to take a bit less.
Never mind that neither side is likely to get ahead for
long in such a race – evolution is blind to the future –
what matters is that evolution can grotesquely exaggerate
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
what initially began as a relatively minor difference. As a
real-world example, female mice that lack a paternally
expressed gene called Peg3 make poor mothers, investing
little effort in their pups (Li et al. 1999).
Genetic imprinting may play a role in human mental
disorders. Badcock and Crespi (2006) suggested that the
“extreme male brain” of autistics is caused by a disrupted
balance between male and female expressed genes. This
fascinating hypothesis could very well explain much
about autism, but a central question remains: Why have
autism susceptibility alleles (alleles that alter the balance
between imprinted genes) persisted in the population?
We discuss two possibilities. Mutation-selection is a potential explanation for this, given the data we reviewed in our
target article that these disorders are associated with
mutation loads (sect. 7). Because mutation-selection maintains substantial variation only in polygenic phenotypes,
this hypothesis requires that many genes are imprinted
and/or regulate imprinted genes.
Gangestad & Yeo provide an alternative possibility:
Alleles that were adapted to a previous counter-adaptation
become maladaptive in the context of a newer counteradaptation. At any given time, some portion of alleles
should be increasing or decreasing in the population,
driven by a futile, but nevertheless consequential, armsrace between maternal and paternal genes. Such constant
flux would maintain some level of maladaptive genetic
variation in the population (Gangestad 2003; Gangestad &
Yeo 1997). In section R2.3, we review some predictions
and evidence that might help distinguish between these
R2.2. Recent selective sweeps
Alleles can be increasing or decreasing in the population
for any number of reasons, not just due to coevolutionary
arms races. Crespi (see also McGrath) suggests that
mental disorders may be, in part, harmful side-effects of
such alleles that have positive or negative net fitness
effects. Of course, it is important to point out that such
alleles cannot have fixated or gone extinct; otherwise, no
genetic variation would result. Human molecular genetics
is revealing several examples of recent selective sweeps
(Cochran et al., in press; Evans et al. 2005), and it is possible that such alleles affect mental disorder risk.
As Crespi acknowledges, “strong selection leads to
maladaptive, more or less transient by-products.” The
question is, how transient is transient? In our view, selective sweeps seem most relevant to explaining populationspecific concentrations of disorders that reflect alleles no
more than a thousand generations (20,000 years) old
(e.g., Cochran et al., in press). Such a time frame would
be consistent with McGrath’s comment: Relative to the
other models (with the exception of G– E interactions),
recent selective sweeps predict variation in prevalence
rates between populations. Selective sweeps that began
further back in time would be increasingly unlikely to
result in polymorphisms today, because even minor
fitness effects drive alleles to fixation or extinction over
such time periods (target article, sect. 4.3). Therefore,
we are highly skeptical of the psychiatric relevance of
selective sweeps that occurred before humans split into
several lineages, such as those concerning the evolution
of schizophrenia, as proposed by T. J. Crow (2000),
Horrobin (2002), or Burns (2004). These earlier selective
sweeps, which ostensibly began before the human
lineage split (50,000 – 200,000 ago), could have had two
effects (see also sect. 5.1 of the target article). First, they
could have brought the susceptibility allele to fixation or
extinction – hence, no genetic variation. Second, the susceptibility allele could have been hitchhiking nearby an
adaptive allele but become unlinked in the course of evolution, in which case the time clock begins when the alleles
became unlinked. Again, if this was much longer than a
thousand generations ago, no genetic variation would
result. Until someone can offer a reasonable explanation
of why 100,000-year-old susceptibility alleles under directional selection are still polymorphic, such “genes that
make us human” hypotheses cannot be considered viable
explanations for the genetic variation in risk of any
mental disorder (Keller 2005).
R2.3. Predictions of non-equilibrium processes
Each of the non-equilibrium models require that previously adaptive alleles that are now sweeping toward
extinction (or currently adaptive alleles sweeping toward
fixation) have side-effects that increase mental disorder
risk. Given that, at some point in their evolution, these
alleles were an adaptive part of a genome that made a
fully functioning individual, it is not clear why this would
often be the case, but it is certainly possible. We need
empirical predictions that can discriminate between equilibrium and non-equilibrium evolutionary processes.
Perhaps the strongest prediction, common to all of the
non-equilibrium evolutionary genetic processes that we
have reviewed, is that susceptibility alleles could be at
any frequency at a given time within a population. This
is similar to the prediction of high-frequency alleles governed by balancing selection, but at a molecular genetic
level, the two processes can be distinguished (Bamshad
& Wooding 2003). Non-equilibrium processes can be
even more easily distinguished from a mutation-selection
balance. Although both might yield predictions that are
similar (the possibility of population-specific susceptibility
alleles), other predictions are quite different (alleles
sweeping to fixation/extinction could be at much higher
frequencies than harmful mutations would reach, and
the sweeping alleles would not cause any of the indicators
of mutation load reviewed in sect. 7). Because they are
more likely to be common, we should expect that alleles
governed by non-equilibrium processes will be relatively
easy to find in gene-mapping studies, although, to date,
there is little evidence that genes of major effect exist for
any of the common mental disorders that are the focus
of our target article. This may simply mean that such
alleles have minor effects, or that they are population
specific. Moreover, some common susceptibility alleles
of importance may yet emerge, and Gangestad & Yeo’s
observation that imprinted genes appear over-represented
among possible susceptibility alleles is provocative.
Of course, as Crespi notes, the most direct evidence in
favor of a selective-sweep hypothesis of mental disorder
alleles would come from (a) finding alleles that are reliably
associated with mental disorders, and (b) demonstrating
that these alleles have been under directional selection.
Crespi cites 20 papers covering 12 different genes in
defense of his assertion that bipolar disorder and
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
schizophrenia susceptibility alleles have been under
positive selection in the human genome. Eleven of these
citations purport to show that certain of these alleles
have been under positive selection, whereas the rest are
primary research publications showing various levels of
support for associations between these alleles and schizophrenia or bipolar disorder. However, if the last 15 years
of gene-mapping studies have taught us anything, it
should be that conclusions based on a handful of association or linkage findings in primary research are nearly
worthless. The chances of Type I errors (the associations
are not real) and Type II errors (many associations were
missed because of small effect sizes) in individual findings
from primary research are almost certain (Sullivan 2005).
Meta-analyses are not free from problems either, but
they are preferable to primary research. We conducted a
PubMed search on each of the 12 genes listed by Crespi
by using keywords “meta-analysis,” the name of the gene
in question, and “schizophrenia” or “bipolar.” If multiple
meta-analyses were found, we report the most recent.
We found no meta-analyses for seven of the genes
SYNJ1), indicating that not enough research had been
accumulated about them to warrant one. The results of
three meta-analyses were negative (for the schizophrenia
links with APOE [Xu et al. 2006] and with DRD4 [Glatt
et al. 2003] and the bipolar link with DRD4 [Lopez
Leon et al. 2005]) and three were positive (for the schizophrenia links with SLC6A4 [Cho et al. 2005] and NRG1
[Li et al. 2006] and for the bipolar link with MAOA
[Preisig et al. 2005]). The effect sizes for the positive findings were quite small (odds-ratios less than 1.25), about
the sizes discussed in our target article (sect. 7.6). These
studies do not provide persuasive evidence that these
genes, individually or together, explain much of the population risk in bipolar disorder or schizophrenia. Nevertheless, meta-analyses are typically conducted by averaging
results that come from different evolutionary lineages,
and it is possible that these genes increase mental disorder
risk in only certain populations, depending upon different
genetic or environmental backgrounds.
R3. Responses to arguments favoring neutral
explanations for mental disorders
R3.1. Are mental disorders beneficial?
Gernsbacher, Dawson, & Mottron [Gernsbacher
et al.] persuasively argue that autism confers specific
cognitive benefits in certain individuals. However, from
an evolutionary genetic perspective, fitness effects of
mental disorder must be defined in terms of evolutionarily
relevant benefits or costs to oneself and ones’ relatives
(inclusive fitness), not just advantages in certain domains
of modern-day functioning (see also sect. R4.2). In fact,
most people with autism have extreme difficulty attracting
and retaining sexual partners, so autism is likely to be evolutionarily harmful. Of course, whether or not a mental
disorder (or any other trait) is evolutionarily maladaptive
says nothing about whether that trait should be considered
“wrong” or “maladaptive,” in the way that the words are
used (interchangeably) in everyday parlance. Adoption of
genetically unrelated offspring may decrease inclusive
fitness, but we hope no one concludes from this that
adoption is wrong!
R3.2. Does the possibility of gene-by-environment
(G –E) interactions suggest that mental disorders
were benign or extremely rare in ancestral
Mayo & Leach, Polimeni, and Wakefield rightly point
out that the paradox of our target article depends on
harmful fitness effects and high prevalence rates of
mental disorders in the ancestral past, a point we also
made in the article (sects. 3.3 and 4). We argued that
the likely ancestral fitness costs of particular disorders
could be inferred to some degree from the current dysfunctions they impose, and the current social and sexual
stigmatization they evoke. If mental disorders existed in
the past as they do today, then at some level of abnormality, they almost certainly lowered ancestral fitness. Of
course, they may not have existed in the past as they do
now. Mayo & Leach critique an argument, attributed to
us, that large G –E interactions were implausible (sect
4.2), and point out that such interactions have been
observed, for example, in diabetes and possibly in
depression. However, we should point out that we have
never doubted that large G– E interactions were possible,
but only that strict neutrality, whereby maladaptive alleles
today were completely neutral ancestrally, was implausible. Nevertheless, Mayo & Leach’s point is perfectly
valid: large G –E interactions cannot be ruled out a priori.
Strictly speaking, therefore, the central paradox of the
target article may not exist, but is this likely? The short
answer is no. We very much doubt Polimeni’s suggestion,
echoing Michel Foucault, that mental disorders may have
been socially unimportant before the eighteenth century.
It would be incredible if people living in ancestral environments were all intelligent, emotionally stable, cheerful,
and sane – in other words, free from mental disorders –
just as it would be incredible if they were all healthy and
attractive. Given that this was not the case, scientists
need good, testable hypotheses for why heritable traits
subjected to continual natural selection nevertheless can
be maladaptive.
Furthermore, writings from ancient sources up through
the Middle Ages describe symptoms consistent with
modern diagnoses of schizophrenia, mental retardation,
depression, bipolar disorder, and anxiety disorders
(Draguns 1982; Jeste et al. 1985; Willerman & Cohen
1990). In modern times, schizophrenia has been found
in investigations of 46 different countries, including
underdeveloped, emerging, and developed countries
(Saha et al. 2005). We did not discuss the historical evidence in our target article because the evidence is contestable, suffering from obvious methodological shortcomings,
but taken together with modern prevalence estimates, it
casts further doubt on suggestions that mental disorders
are purely modern phenomena.
In our target article, we argued that G –E interactions
can be inferred from highly variable mental disorder
prevalence rates over time or location (e.g., 1% versus
20%), plausible interacting environmental agents, and
the possibility of very high prevalence rates. Some
mental disorders, especially depression, follow these
patterns, but the most serious mental disorders – mental
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
retardation, schizophrenia, and bipolar disorder – do not,
or do to a much lesser degree. Here, we offer one
additional empirical statistic that might help weigh
whether G– E interactions explain high prevalence rates
of a given mental disorder: its minimum prevalence rate
across cultures. The minimum rate suggests how low its
rate could have been in ancestral environments, barring
the unlikely event that all cultures worldwide have
already gone through the same relevant environmental
shifts. If reliable estimates of the minimum rate across
cultures approach zero, then ancestral neutrality of susceptibility alleles is a plausible hypothesis. However, so
long as the cross-cultural minimum is non-negligible
(.50 per hundred thousand), its prevalence requires an
Finally, commentators who advocated ancestral neutrality and strong G– E interactions failed to grapple
with some of our key theoretical and empirical objections
to this explanation. For example, ancestral neutrality predicts that any prevalence rates, from 0% to 100% across
mental disorders, should be observed, rather than those
(,5%) typically observed (sect. 4.3). Perhaps most importantly, evidence that inbreeding increases mental disorder
risk is seriously damaging to explanations that disorders
were ancestrally neutral or adaptive, because the direction
in which traits move following inbreeding clarifies the
direction of low fitness in ancestral environments (contra
Sherman, who thinks that inbreeding provides information on modern fitness effects).
For these reasons, we do not think that widespread and
large G– E interactions can fully explain the high prevalence rates of some of the most serious mental disorders,
but we do think it is a plausible model for some of them
(sect. 4.4). What are severely needed, but lacking, are
good data on the prevalence rates of mental disorders
from traditional societies, whose lifestyles are probably
more representative of ancestral conditions. We think
that one of the highest priorities for Darwinian psychiatry
should be to enlist the help of anthropology collaborators
to carry out such research.
an important counterbalance to the prevailing assumption
that all mental disorders are dysfunctional. Indeed, the
authors themselves have proposed adaptive functions for
states considered disorders in mainstream psychiatry –
depressive symptoms in response to losses or threats
(Keller & Nesse 2006), and female vaginismus or anorgasmia in response to low-quality sexual partners (Jenkins &
Miller 2005; Miller 2005). Natural selection should optimize facultative (environmentally responsive) adaptations
which are reliably triggered in all humans in certain situations. However, the identification of adaptive benefits
for a putative disorder does not imply that balancing selection maintains the alleles that cause the genetic variation
of that disorder (see also sects. 1.2 and R4.2).
As Troisi points out, adaptationist hypotheses are easy
to create but hard to falsify, although this problem is not
unique to adaptationist hypotheses (Andrews et al.
2002). Researchers forwarding hypotheses that mental
disorders are adaptive should be eager to find empirical
avenues that would help falsify their favored hypotheses.
We identified several such avenues in our original target
article. Here, we offer another.
Hypotheses that mental disorders are themselves
complex adaptations, maintained by balancing selection
(Polimeni, Price, Sherman), are some of the only
evolutionary hypotheses of mental disorders that are not
consistent with any degree of mutation-selection explaining susceptibility alleles. This is because it is absurdly
improbable that mutations (and other developmental
insults) would lead to full-fledged, complex adaptations
by chance. No commentator has rebutted this point. We
believe that we have made a compelling case that
mutations are responsible for at least some, and we think
much, of the genetic variation in mental disorder risk
(sects. 7 and R5.7). Such evidence is devastating to
hypotheses positing that bipolar disorder, schizophrenia,
autism, and the like, are themselves complex adaptations.
R4. Responses to arguments favoring balancing
selection explanations for mental disorders
Crespi, Klimkeit & Bradshaw, and Nettle point out
that the empirical evidence of links between creativity –
schizotypy and creativity – bipolar disorder are actually
quite robust. We cede the point – there appears to be a
stronger mental disorder –creativity link than we acknowledged – but we disagree that this, or any other supposed
benefit of mental disorders, constitutes sufficient evidence
for balancing selection. If nothing else, we hope that our
target article highlights the additional types of evidence
required to make a case for various types of evolutionary
explanations of genetic polymorphisms. An argument for
balancing selection also entails: (a) explaining why the
net fitness effect of the susceptibility allele is equal to
the fitness effect of the non-susceptibility allele; (b)
explaining why one of these alleles doesn’t fixate by
chance (which probably entails explaining how its fitness
depends on its frequency); and, critically, (c) presenting
data that support predictions from balancing selection
Our point is that simply positing a putative benefit of a
mental disorder is not at all convincing evidence that the
mental disorder is maintained by natural selection.
Not surprisingly, many of the commentators most
critical of our target article were those who have
invested the most in balancing selection arguments
(Klimkeit & Bradshaw, Polimeni, Preti & Miotto,
Price, Sherman, Wilson), which posit overt or hidden
adaptive benefits to certain mental disorders or their susceptibility alleles. Other commentators developed more
nuanced arguments that balancing selection plays a larger
role than we suggested (e.g., Allen & Badcock, Easton,
Schipper, & Shackelford [Easton et al.], Crespi). In
this section, we respond to commentators who argued for
the viability of balancing selection explanations.
R4.1. Are mental disorders complex adaptations?
We welcome several commentators’ efforts to question
whether particular mental disorders are truly harmful
(Gernsbacher et al., Easton et al., Klimkeit &
Bradshaw, Polimeni). As we stated, such questioning is
R4.2. Are demonstrations of benefits among those with
mental disorders or their relatives (e.g., creativity)
sufficient evidence for balancing selection?
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
Creativity may be one advantage (among many disadvantages) of bipolar or schizophrenia susceptibility alleles.
However, genes typically have many effects (pleiotropy).
Mutations or the disorders they cause that have a net
fitness-reducing effect nevertheless may have apparently
beneficial effects in particular domains. For example,
those with Down syndrome have high self-esteem
(Glenn & Cunningham 2001) and are described as being
cheerful, happy, and helpful (Meyers & Pueschel 1991).
If true, would this imply that Down syndrome has been
maintained by balancing selection? Likewise, a congenital
inability to feel pain could yield such benefits as higher
bravery and risk seeking, but its net effect on fitness is
certainly negative because it usually results in early
death (Sternbach 1963). If one looks hard enough, positive
side-effects could be posited or demonstrated for
practically any disorder. Depression saves energy in the
winter, whereas mania boosts it in the spring
(Sherman); schizophrenia alleles increase creativity in
relatives (Crespi), or lead to group dispersal (Price),
or increase the chance that one becomes a shaman
(Polimeni), and so forth.
However, the question is not whether susceptibility
alleles for bipolar disorder, autism, and schizophrenia – or
for a congenital inability to feel pain, or for Down syndrome for that matter – have positive effects in particular
life domains. The question is whether the net fitness effects
of these susceptibility alleles, summed across all relevant
life domains and across all possible genomes that they
might find themselves in, are positive, negative, or (as
required by balancing selection) exactly zero. It seems
likely to us that the benefits of increased creativity do not
compensate for the increased risk of mental disorders,
especially given that there are probably many ways to be
creative without having an increased risk of mental disorders. Of course, this is a plausibility argument, as are
the arguments by commentators who suggested that the
mental disorder –creativity link was evidence of balancing
selection. What predictions distinguish these possibilities?
Balancing selection arguments typically imply strong,
discriminating predictions at the genetic level, which
their advocates failed to discuss. A whole host of evolutionary genetic models predict that balancing selection maintains just a few polymorphic loci (though see sect. 7.6 of
the target article) harboring a few (usually just two)
common alleles at each locus (Barton & Keightley 2002;
Mani et al. 1990; Roff 1997). (Contra Wilson, the
number of common alleles does matter, and provides evidence as to which evolutionary processes explain them.)
Such alleles should be easy to find by using current
gene-hunting methods, and continued lack of progress in
finding them would augur poorly for such a hypothesis
(sect. R5.8). The strong evidence (sects. 7 and R5.7) that
harmful mutations play at least some role (and probably
an important one) in mental disorder risk also casts
doubt on balancing selection being a general resolution
to the paradox.
How, then, is one to reconcile evidence of the mental
disorder –creativity link in light of evidence that mutations
play an important etiological role? One obvious answer is
that no reconciliation is necessary; as we just described,
not every facet of a disorder need be negative for its net
effect to be negative. An alternative explanation is provided by Nettle. The normal range of creativity may be
nearly neutral (or under balancing selection). Most of
the genetic risk of mental disorders comes from harmful
mutations. High creativity has negative effects on fitness
when coupled with a high mutation background because
it increases the risk of mental disorders, but it has positive
effects when in a low mutation background. Similarly,
creativity could be a sexually selected signal, designed to
partially reveal one’s mutation load: only those with a low
mutation load can afford the cost of being creative. By
this view, mental disorders are the result of being too creative in the context of a high mutation load (Shaner et al.
2004). In neither scenario are mental disorders directly
maintained by balancing selection, and they each demonstrate how mutation explanations can be consistent with
certain forms of balancing selection or neutral evolution
acting on other traits. Of course, such explanations also
must explain the variation in creativity or in mismatches,
and both must make empirically tractable predictions
before being accepted.
R4.3. Does increased fertility in relatives of those with
mental disorders indicate heterozygote advantage?
Crespi and Polimeni both say that there is good evidence
for heterozygote advantage in mental disorders. Because
Polimeni provides no support for this, we will focus on
Crespi’s claim, which argues that we were too quick to
dismiss three previous “positive” studies (Avila et al.
2001; Fananás & Bertranpetit 1995; Srinivasan &
Padmavati 1997) that found higher reproductive success
in first-degree relatives of schizophrenics in favor of a
single study (Haukka et al. 2003) that did not. We want
to make four points regarding these studies and their
implications. First, several other studies have also failed
to find an increase in reproductive success among firstdegree relatives of schizophrenics (Bassett et al. 1996;
Buck et al. 1975; Rimmer & Jacobsen 1976; H. P. Vogel
1979). Second, the Haukka et al. (2003) study is important
not only because it was so much larger than previous ones
(11,000 schizophrenics versus fewer than 200 in previous
studies), but, more importantly, because it was the best
controlled of all the studies. The three previous “positive”
studies ascertained schizophrenic patients from psychiatric research hospitals, and (for example) the types of
families that admit their relatives to psychiatric research
hospitals might not be representative of the families of
all schizophrenics. The Haukka et al. (2003) study, on
the other hand, was conducted on hospitalization data
from every individual born in Finland from 1950 to
1959, and thus the chance of unforeseen biases (e.g., the
types of patients ascertained) was negligible. Third,
modern fertility has an unknown relationship to ancestral
fertility (sects. 3.3, 4, and 7.1). More persuasive evidence
for heterozygote advantage would come from data
showing some type of credible benefits among unaffected
relatives, such as creativity, but such evidence would be
only one piece of the puzzle (sect. R4.2).
Last, and echoing our remonstrations from section R4.2,
higher reproductive success in relatives – or any other
putative benefits among them – is not itself evidence for
heterozygote advantage; it is merely one explanation consistent with such a hypothesis. So, too, would Nettle’s or
Shaner et al.’s (2004) hypotheses provide an explanation
for such findings. We need discriminating predictions
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
based on evolutionary genetic theory, not just a collection
of interesting findings that can be cobbled together to
form a plausible story. For example, heterozygote advantage predicts very high dominance variation in particular
(Falconer & Mackay 1996), in addition to high non-additive genetic variation in general. Do we see this for schizophrenia, or for bipolar disorder for that matter? If the
genetic variation in these disorders is due to dominance
variation, parent and offspring risk ratios should be quite
low, whereas sibling risk ratios should be higher. The relative risk for schizophrenia is 6 for parents of schizophrenics and 13 for offspring of schizophrenics, whereas the
sibling risk is 9 (Gottesman 1991). (Parents had the
lowest rate because schizophrenics are less likely than
non-affected individuals to have children [Haverkamp
et al. 1982]). Similarly, no consistent differences between
risks in parents, offspring, and siblings for bipolar disorder
have been observed in the literature (Craddock & Jones
1999). The genetic action involved in these disorders
appears to be due to additive and/or epistatic genetic
action, but probably not to genetic dominance. If one
hypothesizes that heterozygote advantage maintains these
mental disorders, one must explain this inconsistency.
Otherwise, it may be time to jettison the hypothesis.
R4.4. Can antagonistic pleiotropy explain mental
disorder susceptibility alleles?
Several commentators forwarded the idea that normal variants might come together in rare, toxic combinations to
produce mental disorders (Allen & Badcock, Wakefield,
Williams). Wakefield and Allen & Badcock both state that
the Eaves et al. (1990) study – which found higher fertility
among those with (a) high neuroticism and low extraversion and (b) low neuroticism and high extraversion – suggested a mechanism by which susceptibility alleles might
be maintained in the population. Leaving aside the
important issue of whether modern fertility can be used
to infer ancestral fertility (sects. 3.3, 4, and 7.1), these
are arguments for antagonistic pleiotropy of susceptibility
alleles – they have positive fitness effects against some
genetic backgrounds and negative fitness effects against
others. Wakefield is not necessarily wrong, however,
when he says this situation is related to non-additive
genetic variation. In many cases, such as this one, antagonistic pleiotropy and non-additive genetic variation are
completely intertwined. Theoretical work (Curtisinger
et al. 1994; Hedrick 1999; Prout 1999) has shown that
antagonistic pleiotropy maintains genetic variation only if
it is caused by heterozygote advantage at just one or a
few loci, in which case almost all the genetic variation
would be non-additive (dominance) variation. Therefore,
the lack of empirical evidence for heterozygote advantage
(sects. 5.4 and R4.3) is also relevant here.
The example of mental disorders being caused by toxic
combinations of normal variants is instructive, because it
illustrates why previous evolutionary genetic models
have found that antagonistic pleiotropy is unlikely to maintain genetic variation in any trait in any species (see also
sect. 5.5). Consider a hypothetical locus that affects neuroticism: the allele N1 at this locus increases neuroticism
and the allele n1 decreases it. Assume that there are also
lots of alleles at different loci that also increase or decrease
neuroticism (N2 and n2, N3 and n3, etc.) and extraversion
(E1 and e1, E2 and e2, etc.). N1 increases fitness when it
finds itself in a body with lots of e alleles in its DNA,
and decreases fitness when it finds itself in a body with
lots of E alleles in its DNA (this is true irrespective of
the number of other N alleles in the DNA). The opposite
happens for n1.
The question is: Will this process itself maintain both
N1 and n1 in the population? The answer is: It will not.
The reason is that the fitness of N1 is very unlikely to be
equal to the fitness of n1. If there are more E alleles
than e alleles in the population, N1 will, across all the
bodies it finds itself in, have a negative fitness effect on
average. It will go extinct, and n1 will fixate. If there are
more e alleles in the population, n1 will go extinct, and
N1 will fixate. In either case, variation at the locus is not
maintained. The exact same dynamic is also working at
all the other loci harboring the other N/n alleles and
E/e alleles. Eventually, they should all fixate in favor of
one combination or the other: either all people will be
highly neurotic introverts, or they will all be non-neurotic
extraverts. Even in the unlikely situation where the
balance of E/e and N/n alleles is such that the fitness of
N1 and n1 are exactly the same, the mechanism is not
stable to perturbations due to random genetic drift. If E
alleles happen to get a little more common by chance,
N1 decreases, which fuels a further increase in E alleles,
and so forth. Unlike frequency-dependent selection or
heterozygote advantage, antagonistic pleiotropy offers no
homeostatic mechanism that maintains alternative alleles
in the population (Curtisinger et al. 1994; Hedrick 1999;
Prout 1999).
R4.5. Is morbid jealousy an adaptation maintained by
balancing selection?
We place our response to Easton et al.’s commentary separately because it raises an interesting issue. Jealousy is a
facultative response (a response to environmental cues),
and there is good reason to believe that it serves an adaptive mate-guarding function (Buss 2000). Easton et al.’s
balancing selection model sees morbid jealousy as a potentially adaptive extreme of normal sexual jealousy. Before
continuing, we would like to note two caveats regarding
this hypothesis. First, such a hypothesis requires the
types of support outlined in sections R4.2 and
R4.3 – demonstrations of putative benefits are not
enough. Second, it seems likely that one could focus on
the extreme end of any behavioral dimension and ask
why “it” has been maintained in the population. Thus,
we agree with Easton et al. that it is important to understand if morbid jealousy is discrete or part of a continuum.
A slight modification of Easton et al.’s model would
make it parallel to our model of depressive responses:
Some morbid jealousy might be a normal and adaptive
reaction to highly jealousy-provoking situations, whereas
other instances of it might be side-effects of developmental and mutation noise (e.g., Yeo et al. 1999) in the neurodevelopmental system. Here’s where things get
complicated in explaining facultative mechanisms:
Where does the error in the neurodevelopmental system
come from? One answer, given in our target article (sect.
1.2), is that it is a side-effect of environmentally and
genetically induced error that disrupts setting the appropriate threshold for sexual jealousy. Alternatively, this
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
error might not directly affect the jealousy system at all,
but rather might be a reflection of overall mutation load
(P. Andrews, personal communication, April 10, 2006).
Individuals with a high mutation load might have a
reduced ability to retain mates and deter rivals, and
might thereby adaptively lower their jealousy threshold.
This latter explanation would also predict that genetic
inbreeding, paternal age, and environmental perturbations
would increase morbid jealousy – not because of disruptions in the jealousy system, but because of lower
general mate value. Morbid jealousy would be common,
harmful, and heritable because low mate value is
common, harmful, and heritable.
A similar explanation might account for some of the
genetic variation of several other disorders that are
extremes of facultative responses. For example, people
who are less able to function in the world, perhaps
because of a high mutation load, may be more likely to
suffer from normal depressive symptoms. Consistent
with this hypothesis, exposure to stressful life events is
itself heritable, and this appears to contribute to the
genetic liability for depression (Kendler & KarkowskiShurman 1997; Silberg et al. 1999).
Ultimately, however, it should be realized that these two
explanations of the source of error in facultative systems
are alternative descriptions of how developmental errors
and mutations affect mental disorders, but they are not
conflicting. They are important to consider because they
force one to think more deeply about the varied pathways
that connect genes to phenotypes.
R5. Response to concerns over the polygenic
mutation-selection balance model
R5.1. Does the watershed model fail to explain how
different mutations lead to a structured set of
mental disorders?
Whether this concern is a strength or a weakness of our
model depends on how much structure one thinks there
is to psychopathology. Behrendt, Polimeni, Price,
Voracek, Williams, and Wilson suggest or imply that
certain current diagnostic categories are real and distinct.
On the other hand, Airey & Shelton, Allen & Badcock,
Buss, Campbell, Osipova, & Kähkönen [Campbell
et al.], Easton et al., Gangestad & Yeo, Klimkeit &
Bradshaw, McGrath, and Troisi suggest or imply that
current diagnostic categories are dimensional, heterogeneous, and ultimately somewhat arbitrary.
This is the fundamental question in psychiatric nosology
(Krueger & Piasecki 2002; Widiger & Samuel 2005) and a
key issue in Darwinian psychiatry (Troisi & McGuire 2002;
Wakefield 1992): Are mental disorders natural kinds best
described as discrete categories, or are they arbitrarily
divided symptom sets that may have a hierarchical or
dimensional structure? Several recent models of mental
disorders based on multivariate genetic studies or exhaustive symptom-structure studies of community samples
(Clark 2005; Kendler et al. 2003; Krueger & Piasecki
2002) support a hierarchical or dimensional structure of
mental disorder categories. Such findings suggest that
the same gene affects multiple disorders, and that the
same disorder is affect by many genes, and are consistent
with the watershed model.
Gangestad & Yeo’s developmental instability theory
offers a useful clarification of our watershed model.
Mutations and other insults perturb finely tuned developmental processes. Owing to the nature of neurodevelopment, certain processes channel along the same
pathways. The disruption of different developmental processes can thereby lead to common outcomes (see also
McGrath). Our failure to cite their seminal papers (Yeo &
Gangestad 1993; Yeo et al. 1999) in our target article is
like a fish not noticing the water in which it swims. We
have been so influenced by that model, that we overlooked
the obvious: Developmental instability can help explain
how mutations and other developmental errors manifest as
mental disorders.
Behrendt offers a different, though not necessarily
inconsistent, hypothesis for how diverse perturbations of
a system can lead to an apparently singular disorder.
He suggests that schizophrenia might seem like a semicoherent category, not because there is some final
common pathway in the neurogenetic watershed, but
rather because developmental errors undermine individuals’ capacities for normal social interactions. In response
to the complex social dynamics that characterize human
life, vulnerable individuals develop symptoms which we
define as schizophrenia.
R5.2. Are mental disorders too diverse to be explained
by a single evolutionary genetic model?
Several commentators are skeptical of a mutation-selection
model on the grounds of epistemic humility. Preti &
Miotto caution against “a ‘theory of everything’ which
explains everything and nothing” (suggesting that a
mutation-selection model does not generate testable
predictions). Wilson and Polimeni are unconvinced that
one evolutionary genetic mechanism could explain
mental disorder susceptibility alleles. Allen & Badcock
warn that premature commitment to a single model runs
the risk of neglecting equally viable ones.
Epistemic humility is sometimes a virtue, but not
always. It is simply false to claim that a single hypothesis
cannot explain many diverse phenomena. Newton’s
theory of gravitation and Darwin’s theory of natural
selection successfully explained rather broad classes of
phenomena. A mutation-selection model has already
been shown to explain parsimoniously thousands of phenotypically diverse Mendelian disorders (sects. 1.3 and
7.6). There is no good philosophy-of-science reason to
think that it cannot do likewise for the sundry symptoms
of mental disorders. All else being equal, the more a
hypothesis explains, the better. Given the evidence
reviewed in this response and in our target article, we
continue to find the theoretical and empirical arguments
strongest for a mutation-selection explanation for some,
and probably much, of the genetic variation behind most
serious mental disorders.
That said, we never claimed that mutation-selection was
the only mechanism responsible for the evolutionary persistence of susceptibility alleles, or that balancing selection
played no role in maintaining them (this was explicitly
stated in sects. 4.4, 5.8, 7.6, and 8). Mutation-selection
explanations for susceptibility alleles are perfectly consistent
with other evolutionary processes – near-ancestral neutrality, maladaptive by-products of balancing selection,
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
coevolutionary arms-races, selective sweeps, and so forth.
We fully expect that each of these processes will explain
the existence of some portion of susceptibility alleles.
We also agree that, in future research, it may sometimes
be best to compare evolutionary genetic models one disorder at a time, given the different prevalence rates, heritabilities, genetic correlations, environmental triggers,
symptoms, and fitness effects of different disorders.
However, we worry that disorder-by-disorder research
reifies DSM or ICD categories in ways that have inhibited
big-picture thinking in Darwinian psychiatry. In particular, researchers who focus on one disorder at a time tend
to miss the genetic correlations and phenotypic comorbidities among disorders that may be generated by a polygenic mutation-selection balance, or that may reflect
balancing selection on general personality traits.
R5.3. Is the mutation-selection explanation pessimistic?
Brüne and Airey & Shelton worry that, to the degree
that mutation-selection explains mental disorder risk, the
search for susceptibility alleles could be a fruitless endeavor. It is true that evidence for a major role of mutations in
the etiology of mental disorders should cause dismay to
those psychiatric gene-hunters who expect that susceptibility alleles have large effects that replicate across populations. However, even if most of the mental disorder
genetic variation is caused by mutation-selection (which
is far from certain), this would not mean that psychiatric
gene hunting is doomed (sects. 7.6 and 8). It would,
however, mean that the quest to find susceptibility
alleles will be much slower than originally anticipated,
and that alternative gene-hunting methods will need to be
implemented (Wright et al. 2003). For example, Airey &
Shelton, Campbell et al., and McGrath argue that
empirical support for a mutation model has made
current drives to simplify the genetic structure of mental
disorders (through finding and using endophenotypes)
all the more imperative. Similarly, Campbell et al. argue
that small regional subisolate populations could be a
gene-hunting gold mine, because each population may
contain different susceptibility alleles at fairly high prevalence rates due to bottlenecks and genetic drift.
R5.4. Is the mutation-selection explanation
Some commentators have drawn a distinction between
adaptationist approaches to understanding susceptibility
alleles (by which they seem to mean balancing selection
models) and a mutation-selection model. We think this is
a false distinction. We are ourselves ardent adaptationists
who fully appreciate the value of evolutionary psychology
in characterizing species-typical psychological adaptations
(e.g., see Keller & Nesse 2006; Miller 2005). We agree
with Brüne and Troisi that evolutionary insights are
crucial for understanding mental disorders, which must
be understood against the background of normally functioning psychological adaptations. As an example, Troisi
has forwarded that lifetime prevalence rates of mental
disorders that are approaching 50% suggest that current
definitions fail to distinguish true disorders from adaptive
responses to difficult situations. Current psychiatric
definitions are driven by corporate interests for profit, as
well as individual interests to not suffer, but evolutionarily
informed definitions would be much more scientifically
useful (Troisi & McGuire 2002; Wakefield 1992).
As Troisi skillfully points out, our target article was not
arguing against adaptationism as a general research strategy. It was arguing against a particular type of Darwinian
psychiatry that immediately reaches into the adaptationist
toolbox to understand genetic variation – usually this
means either balancing selection, or, worse (because it
doesn’t explain adaptive genetic polymorphisms), directional selection. Our mutation-selection watershed
model is indeed not adaptationist, but neither is it antiadaptationist. Our model is perfectly compatible with
adaptationism; indeed, for it to have much explanatory
power, it requires the extraordinary complexity and polygenic basis of biological adaptations. If human consciousness could arise from a simple brain built by a small set of
genes, it would be wonderfully mutation-resistant. Alas,
we think human consciousness can only arise in a brain
of gargantuan complexity built by thousands of genes
and that has a vast mutation target size. In our view,
mental disorders are side-effects of adaptive complexity,
not arguments against adaptive complexity.
We think that balancing selection is conceptually
attractive to some commentators because it promises to
bridge the gap between human universals (as studied by
evolutionary psychology) and individual differences (as
studied by personality researchers and psychiatrists) –
and to do so within a behavioral ecology framework that
embraces optimality modeling and evolutionary game
theory. Mutation-selection models are anathema to a
narrow view of behavioral ecology because they undermine the rational-agent assumptions of optimality modeling. In a perfect world with no phenotypic disruptions
and no behavioral errors, balancing selection is the only
force that can make sense of individual differences. Fundamentally, our point is that adaptationists do not have
to believe that we live in this perfect world. Mutations
and other developmental errors happen, and they mess
us up.
This is absolutely not to say that most Darwinian psychiatry has been a waste of time. We are convinced, for
example, that depression has some close relationship to
normal sadness, grief, and transient low mood, and that
these states have ancestrally adaptive functions (Keller &
Nesse 2006), as several Darwinian psychiatrists have theorized before (Gilbert 1992; Hagen 1999; Nesse 2000; Price
et al. 1994; Watson & Andrews 2002). Our question is: Are
these adaptive mood states – complex, finely tuned, and
polygenic as they are claimed to be – immune to disruption? The same question applies to the many plausible
hypotheses about adaptive functions of personality
disorders, sexual disorders, and so forth. We can probably
keep many of the insights from such work, as long as we
bear in mind the likelihood that any complex psychological
adaptation, by necessity, is susceptible to breakdowns
induced by environmental and genetic insults.
R5.5. Do mutations during spermatogenesis fail to
explain the paternal age effects observed for
Crespi has cited five studies (Farrer et al. 1992;
Malaspina et al. 2001; Reik et al. 1993; Sipos et al. 2004;
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
Tiemann-Boege et al. 2002) that, he says, show that
genomic imprinting effects provide a persuasive alternative hypothesis for paternal age effects on schizophrenia
risk. However, three of these citations (Farrer et al.
1992; Reik et al. 1993; Tiemann-Boege et al. 2002)
regard Huntington’s disease and achondroplasia, Mendelian disorders with much simpler modes of inheritance
than any complex mental disorder. These studies are
therefore not relevant to judging whether the paternal
age effects in complex mental disorders have anything to
do with genomic imprinting. On the other hand, the
Sipos et al. (2004) and Malaspina et al. (2001) studies
are indeed relevant to the issue, but they support a mutational explanation over an imprinting one. Malaspina et al.
(2001) stated only that genomic imprinting might explain
the effect, not that it was likely. Malaspina later followed
up her study in order to test which explanation was
correct, and concluded that mutations, not genetic
imprinting, explained the paternal age effects, because
the effect was due solely to sporadic (non-familial) cases
(Malaspina et al. 2002). Similarly, in a better-controlled,
population-based cohort study, Sipos et al. (2004) also
found that the paternal age –schizophrenia link existed
only among sporadic cases. Contra Crespi, the evidence
strongly supports the hypothesis that the link between
paternal age and schizophrenia (and perhaps other
mental disorders [sect. 7.3]) exists because older fathers
are more likely to pass on new, deleterious mutations to
their offspring, who are thereby more likely to develop
schizophrenia, depression, anxiety disorders, and mental
retardation (Bassett et al. 2000; Blackwood et al. 2001;
Inoue & Lupski 2003; MacIntyre et al. 2003). Contra
Wilson, there is no theoretically important distinction to
be made between such chromosomal abnormalities and
point mutations – they are all different types of mutations
that affect the phenotype because they disrupt proper
gene functioning. Of course, chromosomal abnormalities
tend to disrupt many genes at once, and so are phenotypically easier to detect and study in pedigree studies. The
fact that major mutations (chromosomal abnormalities)
can cause mental disorders strongly suggests that minor
ones do, too. An additional piece of evidence that we
have learned more about since writing the target article
is that ionizing radiation, which can cause both
germ-line and somatic mutations, increases the risk of
mental retardation and schizophrenia (Loganovsky &
Loganovskaja 2000; Otake 1996). Both pieces of evidence
strongly implicate a role for harmful mutations in mental
disorder risk. A perusal of the results from an academic
search engine (e.g., PubMed) with keywords “schizophrenia,” “bipolar disorder,” “autism,” or “mental retardation” and “inversion,” “translocation,” “chromosomal
abnormality,” or “ionizing radiation” should make sober
reading for anyone who posits that these disorders are
themselves adaptations (sect. R4.1).
R5.6. Are mutation rates adaptively modulated, and
does this imply that mutations are not harmful?
Although this issue was not raised by any reviewer or
commentator, we end this section on mutation-selection
by amending a point we made in our target article. In
section 7.6, we stated that to the degree mutationselection explains mental disorder susceptibility alleles,
the common disease, rare variant (CDRV) hypothesis
should be true, and if so, mental disorder susceptibility
alleles will be difficult to detect using modern gene
hunting methods. Both statements are correct. We followed this, however, by stating that the current lack of
progress in finding genes made the CDRV more likely.
Here we probably overstated our case. While the lack
of progress in finding mental disorder susceptibility
alleles does suggest that mental disorders are highly polygenic (itself a prediction of the mutation-selection
hypothesis), it does not necessarily imply that the susceptibility alleles at these polygenic loci are individually rare
(i.e., that the CDRV hypothesis is true). The reason is that
most published gene hunting studies have used two
approaches (linkage and candidate gene association)
that provide little and/or imperfect information on the
validity of the CDRV hypothesis. Whole genome association tests, on the other hand, can provide such information, but few such tests have been conducted to
date. This state of affairs is set to change over the next
five years. Instead of being framed as corroborating
evidence, therefore, our point in this section should
have been framed as a prediction: to the degree that
mutation-selection explains the existence of mental disorder susceptibility alleles, forthcoming whole genome
association tests will not be the silver bullet that they
are hoped to be.
Williams appears to suggest that, because mutation rates
are adaptively higher in certain restricted regions of the
genome (Metzgar & Wills 2000), the consequent entropy
stemming from them is not generally deleterious.
However, no evolutionary geneticist that we are aware
of subscribes to the notion that mutations are not generally deleterious, or that mutation loads are not real.
Besides the extensive evidence from mutation-accumulation
experiments that species do suffer from mutation loads
(e.g., see Garcı́a-Dorado et al. 2004), it should be
proof enough to note the extraordinary adaptations
within cell nuclei that minimize the rate or effects of
mutations: copying enzymes, proofreading enzymes,
repair enzymes, cell suicide mechanisms; indeed, DNA
itself is probably an improvement over its progenitor,
RNA, because DNA is less likely to mutate (Ridley
2000). It is true that mutation rates differ across the
genome, and some of this difference is probably
adaptive, but this only signifies that, in certain areas of
the genome where molecular variation is adaptive (e.g.,
the MHC region), selection against mutations is much
R5.7. Are chromosomal abnormalities fundamentally
different from mutations?
As briefly mentioned in the target article (sect. 8), chromosomal abnormalities (translocations, inversions, and so
forth) sometimes cause behavioral syndromes that are
otherwise identical to bipolar disorder, autism,
R5.8. Is the lack of progress in finding susceptibility
alleles evidence for the common disease, rare variant
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
R6. Responses to particular points raised by
R6.1. Do mental disorders require different explanations
than other types of disorders?
Mayo & Leach argue that the explanations we provide for
understanding the evolutionary maintenance of genetic
variation are not unique to mental disorders, and that
other types of disorders, such as diabetes, asthma, heart
disease, and endometriosis, can be understood by using
these same principles. We completely agree with this
important point (briefly alluded to in sects. 2 and 8). The
distinction between mental and physical disorders is,
from a biological perspective, an arbitrary one. We think
that the polygenic mutation-selection model applies to
most organic dysfunction across all species, as many biologists have argued before us. It was a pragmatic decision
not to extend our discussion to other harmful, heritable
physical disorders or to other species.
R6.2. Do Keller & Miller understate the role of the
Allen & Badcock assert that our target article
understated the role of complex G– E interactions in
phenotypic development. We agree that normal human
development depends on complex interactions among
genes, organisms, and the environment to produce
psychological adaptations. However, in referring to the
study by Caspi et al. (2003), Allen & Badcock (see also
Voracek) stated that strong G –E interactions are the
rule rather than the exception. Care must be taken in
placing such findings in the proper context. G– E interactions that are relevant to understanding the prevalence
of mental disorders are those with respect to ancestral
versus modern environments, not necessarily with
respect to the environmental factors (stressful life
events) studied by Caspi et al. (2003). If the environmental
factors identified in such G– E interaction research were
also present in ancestral environments, even if only
occasionally, then the susceptibility alleles would still
require explanation.
R6.3. Does it make a difference whether alternative
adaptive strategies are environmentally or genetically
Price makes the point that behavioral ecology is an
important tool for understanding psychiatry (we agree),
but he also states that it makes no difference whether
alternative adaptive strategies are environmentally or
genetically determined, which is a point we emphatically
disagree with. It simply does make a difference. If alternative strategies are environmentally determined, such that
organisms adaptively shift strategies depending on the
situation, the genes affecting this shifting behavior would
be at fixation. In this case, straightforward directional
selection arguments apply (the fittest alleles fixate in the
population). If an adaptive behavioral polymorphism is
genetically determined, on the other hand, then the only
mechanism that could explain its existence is balancing
selection, which we discussed in detail, and which makes
specific, testable predictions. A central point in our
target article was that the type of process used to explain
most adaptations – directional selection – cannot explain
an adaptive genetic polymorphism (sects. 5.1 and R2.2).
R6.4. Did Keller & Miller misrepresent the work of
We feel impelled to reply to Wilson’s statement that we
engaged in ad hominem attacks and misrepresented his
work. Clearly, we intended neither of these. The “wouldbe ad hominem” that Wilson describes was intended to
draw attention to the fact that many psychiatrists, both
researchers and clinicians, hold a dim view of what they
consider to be Panglossian perspectives of mental disorders held by many Darwinian psychiatrists (see also
Troisi’s point in this regard). We did not mean to imply
that Darwinian psychiatrists are “befogged in the Ivory
Towers.” Indeed, we consider ourselves ardent Darwinian
psychiatrists, and we freely admit to having no experience
working in psychiatric hospitals. Our wording should have
been more careful.
We disagree, however, that we misrepresented the
papers of Wilson that we cited, which were forceful
deductive arguments that balancing selection maintains
bipolar disorder, depression, and certain other common
mental disorders. For example, Wilson (1998, p. 381)
states that,
Several important psychiatric conditions (manic depression,
sociopathy, obsessive-compulsivity, anxiety, and even drug
abuse and some disorders of personality) are so common and
so strongly epigenetic that their epidemiological frequencies
surpass even quite conservative thresholds of evolutionary
selection. The deduction follows that such frequency
thresholds were surpassed due to the Darwinian selection of
genes advantageous over the course of evolution.
Wilson goes on to demonstrate how much more
common bipolar disorder is than expected under a
single-gene mutation-selection model, with the implication being that the disorder is so preposterously
prevalent compared to what it would be under a
mutation-selection balance that it must be maintained by
natural selection (Wilson 1998, p. 390). Similarly, in a
letter to the editor of Archives of General Psychiatry, he
states, “It is clear . . . that some underlying genotypy of
depression is altogether too common to not have been
selected” (Wilson 2001, p. 1086). It seemed clear to us
from these writings that Wilson portrayed balancing
selection as the only viable option for explaining the very
high prevalence rates of bipolar disorder, depression,
and certain other mental disorders.
R6.5. Sex differences in mental disorders require
We agree with Brüne that Darwinian psychiatry can and
should help explain why males and females differ in
the way mental disorders are manifested. Although a
mutation-selection model alone does not explain sex
differences in mental disorders, it can be easily integrated
into an adaptationist framework to do so. We expect that
the neurogenetic watershed will be sex differentiated,
because human brain development and behavioral development are sex differentiated. The effects of mutation
load will be overlaid on the background of a normal
Response/Keller & Miller: Resolving the paradox of heritable mental disorders
psychology, and mutations that disrupt the sex-specific
watersheds will tend to have sex-specific effects. The
result may be that, for example, some of the same
mutations that increase the risk of sociopathy in males
increase the risk of borderline personality disorder in
R6.6. Does mutation-selection explain the genetic
variation in personality?
We find it more likely that balancing selection explains the
genetic variation in personality traits than that it explains
the genetic variation in mental disorder risk. Personality
traits are relatively orthogonal to one another, are not
associated with the same levels of dysfunction as mental
disorders, and have plausibly balanced fitness costs and
benefits. Several personality traits are positively correlated
with indexes of genetic or phenotypic quality (such as
intelligence, health, attractiveness, and body symmetry),
but these tend to be much lower in magnitude than
these relationships among mental disorders. In these
respects, personality traits are quite different from
mental disorders, and are better candidates for balancing
selection explanations.
As Buss pointed out, however, one could make the
opposite argument: If mutation-selection explains most
mental disorder susceptibility alleles, and if mental disorders are continuous with normal personality variation
(Benjamin et al. 2002), then mutation-selection should
explain the genetic variation of normal personality, as
well. This viewpoint has intuitive appeal: Everyone
harbors hundreds of deleterious mutations, but most
people never develop a mental disorder. How are all
those susceptibility alleles expressed in the nondisordered population? In support of this view, Buss
noted that one end of each personality dimension is
usually more socially and sexually attractive than the
other end (e.g., openness vs. rigidity, emotional stability
vs. neuroticism, etc.). Buss also emphasized other
aspects of personality traits that are consistent with a
mutation-selection model, including their non-additive
genetic variance; their mild assortative mating coefficients;
their social, sexual, and reproductive payoffs; and
their salience in person perception as potential
indicators of mutation load. We add that the lack
of progress in finding genes of major effect in
personality (Munafo et al. 2003) is also consistent with
this hypothesis.
We offer two possible addenda to Buss’s insights.
First, in discussing the mutation-selection model of
mental disorders, we implicitly focused on directional
selection: Selection should disfavor alleles that increase
mental disorder risk. But mutation-selection models
can also maintain variation in traits under stabilizing
selection, and it seems possible to us that some midpoint
in personality is optimal, with both extremes being
suboptimal (perhaps leading to very different types of
mental disorders). This doesn’t seem to square with
evidence that Buss discussed that one end of each
dimension is more attractive than the other; but it is
possible that the population average is to the left or
the right of the most-fit midpoint, such that a linear
effect would exist in addition to a quadratic one in
personality attractiveness. Second, mutations might
affect personality directly, as disruptions in behavioral
mechanisms, but they might also do so indirectly if personality levels are facultative and adaptive (see also sect.
R4.5). For example, if extraversion levels are adaptively
set based, in part, on self-perceived mating quality
(say), then extraversion would indeed be affected by
mutations, but indirectly, because mutations degrade
mating quality.
In any case, the variation in personality must be
explained, and along with Buss, we find a mutational
explanation plausible, if not likely. If mutations explain
the genetic variation in personality, we should expect
that personality (perhaps measured in absolute deviation
units from the mean, if it is under stabilizing selection) is
associated with all the empirical consequences of
mutation-selection (sect. 7).
Wakefield was skeptical of any viewpoint that
would “reduce much normal variation to mild biological
dysfunction,” but we do not see a priori why this is
unlikely. Evidence from mutation-selection studies in
biology suggests that maladaptation is ubiquitous, not
just at the extreme ends of the distributions, but everywhere. Much of what we consider “normal” variation
may indeed be mild dysfunction, even if too mild to
be considered dysfunctional by any useful diagnostic
R7. Conclusion
Humans and other long-lived species are beset by mutational noise. Given the enormous complexity of human
behavioral adaptations, it should come as no surprise
that this maladaptive noise expresses itself behaviorally.
We have argued that this provides a compelling explanation for the existence of some portion of mental disorder
susceptibility alleles. A mutation-selection hypothesis for
susceptibility alleles is not only testable, but a wealth of
already available empirical data strongly support its role
in maintaining some, and perhaps much, of the genetic
risk of several common mental disorders. Nevertheless,
as we have stressed throughout, such evidence does not
preclude the roles of other evolutionary genetic
Each of the evolutionary processes we have reviewed
makes discriminating predictions at the phenotypic,
macro-genetic, and molecular genetic levels. The
testing of different evolutionary genetic models is accelerating. Gene mapping is revolutionizing this endeavor
because it provides a physical measure that partially
reveals the history of genes, weighing in on the type of
evolutionary process responsible for a given genetic
polymorphism (Bamshad & Wooding 2003). McGrath
says, “Let the data rule,” and we could not agree
more. It is exciting, and humbling, to realize that, within
the next 50 years, we will probably know why alleles
that predispose people to common, harmful, heritable mental disorders have persisted over evolutionary
For helpful guidance and generous feedback, we thank Paul
Andrews, Rosalind Arden, Steven Gangestad, Michael Neale,
Brien Riley, and Kim Tremblay.
References/Keller & Miller: Resolving the paradox of heritable mental disorders
[Letters “a” and “r” appearing before authors’ initials refer to target article
and response, respectively.]
Abaskuliev, A. A. & Skoblo, G. V. (1975) Inbreeding, endogamy and exogamy
among relatives of schizophrenia patients. Genetika 11:145 –48. [aMCK]
Abdolmaleky, H. M., Thiagalingam, S. & Wilcox, M. (2005) Genetics and epigenetics in major psychiatric disorders: Dilemmas, achievements, applications,
and future scope. American Journal of Pharmacogenomics 5:149– 60. [BC]
Abraham, A., Windmann, S., Daum, I. & Gunturkun, O. (2005) Conceptual
expansion and creative imagery as a function of psychoticism. Consciousness
and Cognition 14:520 – 34. [BC]
Airey, D. C., Castillo-Juarez, H., Casella, G., Pollak, E. J. & DeVoogd, T. J. (2000)
Variation in the volume of zebra finch song control nuclei is heritable:
Developmental and evolutionary implications. Proceedings of the Royal Society
of London, series B: Biological Sciences 267(1457):2099 –104. [DCA]
Akin, D., Manier, D. H., Sanders-Bush, E. & Shelton, R. C. (2004) Decreased
serotonin 5-HT2A receptor-stimulated phosphoinositide signaling in fibroblasts from melancholic depressed patients. Neuropsychopharmacology
29(11):2081– 87. [DCA]
(2005) Signal transduction abnormalities in melancholic depression. International Journal of Neuropsychopharmacology 8(1):5– 16. [DCA]
Allen, J. S. & Sarich, V. M. (1988) Schizophrenia in an evolutionary perspective.
Perspectives in Biological Medicine 32:132 – 53. [aMCK]
Allen, N. B. & Badcock, P. B. T. (2003) The social risk hypothesis of depressed
mood: Evolutionary, psychosocial, and neurobiological perspectives. Psychological Bulletin 129(6):887 – 913. [NBA]
Allison, A. C. (1954) Notes on sickle-cell polymorphism. Annals of Human Genetics
19:39 –51. [aMCK]
American Psychiatric Association (1994) Diagnostic and statistical manual of mental
disorders, 4th edition. American Psychiatric Association. [MAG]
(2000) Diagnostic and statistical manual of mental disorders, 4th edition (Text
Revision). American Psychiatric Association. [JAE]
Anastasi, A. & Foley, J. P. (1949) Differential psychology. Macmillan. [JAS]
Andolfatto, P. (2001) Adaptive hitchhiking effects on genome variability. Current
Opinion in Genetics and Development 11:635 – 41. [BC]
Andreasen, N. C. (1987) Creativity and mental illness: Prevalence rates in writers
and their first degree relatives. American Journal of Psychiatry 151:1650 – 56.
(2005) The creating brain: The neuroscience of genius. Dana Press. [EIK]
Andres, A. M., Soldevila, M., Navarro, A., Kidd, K. K., Oliva, B. & Bertranpetit, J.
(2004) Positive selection in MAOA gene is human exclusive: Determination of
the putative amino acid change selected in the human lineage. Human Genetics 115:377– 86. [BC]
Andrews, P., Gangestad, S. W. & Matthews, D. (2002) Adaptationism: How to carry
out an exaptationist program. Behavioral and Brain Sciences 25:489– 553.
[rMCK, AT]
Arajärvi, R., Haukka, J., Varilo, T., Suokas, J., Juvonen, H., Suvisaari, J.,
Muhonen, M., Suominen, K., Tuulio-Henriksson, A., Schreck, M., Hovatta, I.,
Partonen, T. & Lönnqvist, J. (2004) Clinical phenotype of schizophrenia in a
Finnish isolate. Schizophrenia Research 67:195 – 205. [TC]
Armus, H. L. (2001) Effect of response effort on the reward value of distinctively
flavored food pellets. Psychology Reports 88(3, Pt 2):1031 – 34. [JW]
Arnold, K. E. & Owens, I. P. F. (1998) Cooperative breeding in birds: A
comparative test of the life history hypothesis. Proceedings of the Royal
Society of London B 265:739 – 45. [AP]
Auranen, M. (2002) Molecular genetics of autism spectrum disorders in the Finnish
population. Doctoral dissertation, Department of Molecular Medicine,
National Public Health Institute and Department of Medical Genetics,
University of Helsinki, Helsinki, Finland. [TC]
Auroux, M. R., Mayaux, M. J. & Guihard-Moscato, M. L. (1989) Paternal age and
mental functions of progeny in man. Human Reproduction 4:794– 97. [aMCK]
Avila, M., Thaker, G. & Adami, H. (2001) Genetic epidemiology and
schizophrenia: A study of reproductive fitness. Schizophrenia Research
47:233 – 41. [rMCK]
Axelrod, R. & Hamilton, W. D. (1981) The evolution of cooperation. Science
211:1390 – 96. [AP]
Badcock, C. & Crespi, B. (2006) Imbalanced genomic imprinting in brain
development: An evolutionary basis for the aetiology of autism. Journal of
Evolutionary Biology 19:1007 –32. [BC, rMCK]
Bah, J., Quach, H., Ebstein, R. P., Segman, R. H., Melke, J., Jamain, S., Rietschel,
M., Modai, I., Kanas, K., Karni, O., Lerer, B., Gourion, D., Krebs, M. O.,
Etain, B., Schurhoff, F., Szoke, A., Leboyer, M. & Bourgeron, T. (2004)
Maternal transmission disequilibrium of the glutamate receptor GRIK2 in
schizophrenia. Neuroreport 15:1987– 991. [SWG]
Bajema, C. J. (1963) Estimation of the direction and intensity of natural selection in
relation to human intelligence by means of the intrinsic rate of natural
increase. Eugenics Quarterly 10:175 – 85. [aMCK]
Balciuniene, J., Syvanen, A. C., McLeod, H. L., Pettersson, U. & Jazin, E. E. (2001)
The geographic distribution of monoamine oxidase haplotypes supports a
bottleneck during the dispersion of modern humans from Africa. Journal of
Molecular Evolution 52:157 – 63. [BC]
Baldessarini, R. J. & Hennen, J. (2004) Genetics of suicide: An overview. Harvard
Review of Psychiatry 12:1– 13. [MV]
Bamshad, M. & Wooding, S. P. (2003) Signatures of natural selection in the human
genome. Nature Reviews Genetics 4:99 – 111. [arMCK]
Bancroft, J. (2002) The medicalization of female sexual dysfunction: The need for
caution. Archives of Sexual Behavior 31:451 – 55. [AT]
Bandelow, B., Krause, J., Wedekind, D., Broocks, A., Hajak, G. & Ruther, E. (2005)
Early traumatic life events, parental attitudes, family history, and birth risk
factors in patients with borderline personality disorder and healthy controls.
Psychiatry Research 134(2):169 –79. [JP]
Barkow, J., Cosmides, L. & Tooby, J., eds. (1992) The adapted mind. Oxford
University Press. [aMCK]
Baron, M., Risch, N. & Mendlewicz, J. (1982) Differential fertility in bipolar
affective illness. Journal of Affective Disorders 4:103– 12. [aMCK]
Baron-Cohen, S. & Belmonte, M. K. (2005) Autism: A window onto the
development of the social and the analytic brain. Annual Reviews of
Neuroscience 28:109 – 26. [EIK]
Barrantes-Vidal, N. (2004) Creativity and madness revisited from current
psychological perspectives. Journal of Consciousness Studies 11:58 – 78.
Barton, N. H. (1990) Pleiotropic models of quantitative variation. Genetics
124:773 –82. [aMCK]
Barton, N. H. & Keightley, P. D. (2002) Understanding quantitative genetic
variation. Nature Reviews Genetics 3:11 – 21. [arMCK]
Bassett, A. S., Bury, A., Hodgkinson, K. A. & Honer, W. G. (1996) Reproductive
fitness in familial schizophrenia. Schizophrenia Research 21:151 – 60. [BC,
Bassett, A. S., Chow, E. W. C. & Weksberg, R. (2000) Chromosomal abnormalities
and schizophrenia. American Journal of Medical Genetics 97:45 – 51. [rMCK]
Behrendt, R. P. (2006) The desire to obtain money: A culturally ritualised
expression of the aggressive instinct (Commentary). Behavioral and Brain
Sciences 29:178 – 79. [R-PB]
Behrendt, R. P. & Young, C. (2004) Hallucinations in schizophrenia, sensory
impairment, and brain disease: A unifying model. Behavioral and Brain
Sciences 27(6):771– 87. [R-PB]
Belsky, J., Steinberg, L. & Draper, P. (1991) Childhood experience, interpersonal
development, and reproductive strategy: An evolutionary theory of
socialization. Child Development 62:647 – 70. [MB]
Benjamin, J., Ebstein, R. P. & Belmaker, R. H., eds. (2002) Molecular genetics and
the human personality. American Psychiatric. [rMCK]
Benson, M. A., Sillitoe, R. V. & Blake, D. J. (2004) Schizophrenia genetics:
Dysbindin under the microscope. Trends in Neurosciences 27(9):516– 19.
Bentley, P. J. (2004) Evolving beyond perfection: An investigation of the effects of
long-term evolution on fractal gene regulatory networks. Biosystems 76(1 –
3):291– 301. [JJM]
Bertolino, A., Blasi, G., Latorre, V., Rubino, V., Rampino, A., Sinibaldi, L., Caforio,
G., Petruzzella, V., Pizzuti, A., Scarabino, T., Nardini, M., Weinberger, D. R. &
Dallapiccola, B. (2006) Additive effects of genetic variation in dopamine
regulating genes on working memory cortical activity in human brain. Journal
of Neuroscience 26(15):3918 – 22. [DCA]
Bertolino, A., Caforio, G., Blasi, G., De Candia, M., Latorre, V., Petruzzella, V.,
Altamura, M., Nappi, G., Papa, S., Callicott, J. H., Mattay, V. S., Bellomo, A.,
Scarabino, T., Weinberger, D. R. & Nardini, M. (2004) Interaction of COMT
Val(108/158) met genotype and olanzapine treatment on prefrontal cortical
function in patients with schizophrenia. American Journal of Psychiatry
161:1798 – 805. [TC]
Beversdorf, D. Q., Smith, B. W., Crucian, G. P., Anderson, J. M., Keillor, J. M.,
Barrett, A. M., Hughes, J. D., Felopulos, G. J., Bauman, M. L., Nadeau, S. E. &
Heilman, K. M. (2000) Increased discrimination of “false memories” in autism
spectrum disorder. Proceedings of the National Academy of Sciences USA
97:8734 –37. [MAG]
Birchwood, M., Meaden, A., Trower, P., Gilbert, P. & Plaistow, J. (2000) The power
and omnipotence of voices: Subordination and entrapment by voices and
significant others. Psychological Medicine 30:337 – 44. [R-PB]
Blackwood, D. R., Visscher, P. M. & Muir, W. J. (2001) Genetic studies of bipolar
affective disorder in large families. British Journal of Psychiatry 178:134 –36.
Bleuler, E. (1911) Dementia Praecox oder Gruppe der Schizophrenien. In:
Handbuch der psychiatrie, ed. G. Aschaffenburg. Franz Deuticke. [aMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Bodmer, W. (1999) Familial adenomatous polyposis (FAP) and its gene, APC.
Cytogenetics and Cell Genetics 86:99 –104. [OM]
Bondy, B., Buettner, A. & Zill, P. (2006) Genetics of suicide. Molecular Psychiatry
11:336 – 51. [MV]
Boone, J. L. (1992) Competition, conflict, and development of social hierarchies. In:
Evolutionary ecology and human behavior, ed. E. A. Smith & B. Winterhalder.
Aldine de Gruyter. [JSP]
Borglum, A. D., Kirov, G., Craddock, N., Mors, O., Muir, W., Murray, V.,
McKee, I., Collier, D. A., Ewald, H., Owen, M. J., Blackwood, D. &
Kruse, T. A. (2003) Possible parent-of-origin effect of dopa decarboxylase
in susceptibility to bipolar affective disorder. American Journal of Medical
Genetics, B: Neuropsychiatric Genetics 117:18– 22. [SWG]
Bostwick, J. M. & Pankratz, V. S. (2000) Affective disorders and suicide risk: A
reexamination. American Journal of Psychiatry 157:1925– 32. [MV]
Bouchard, T. J., Lykken, D. T., McGue, M., Segal, N. & Tellegen, A. (1990) Sources
of human psychological differences: The Minnesota study of twins reared
apart. Science 250:223 – 28. [aMCK]
Brabban, A. & Turkington, D. (2002) The search for meaning: Detecting
congruence between life events, underlying schema and psychotic symptoms.
In: A casebook of cognitive therapy for psychosis, ed. P. Morrison. BrunnerRoutledge. [JW]
Bradshaw, J. L. (2001) Developmental disorders of the frontostriatal system:
Neuropsychological, neuropsychiatric, and evolutionary perspectives.
Psychology Press. [EIK]
Brent, D. A. & Mann, J. J. (2005) Family genetic studies, suicide, and suicidal
behavior. American Journal of Medical Genetics, C: Seminars in Medical
Genetics 133:13 –24. [MV]
Brooks, R., Bussiere, L. F., Jennions, M. D. & Hunt, J. (2004) Sinister strategies
succeed at the cricket World Cup. Proceedings of the Royal Society of London,
B: Biological Sciences 271(Suppl 3):S64– S66. Available at http://¼pubmed&cmd¼Retrieve&
docsum [DRW]
Brown, A. S., Schaefer, C. A., Wyatt, R. J., Begg, M. D., Goetz, R. R., Bresnahan,
M. A., Harkavy-Friedman, J., Gorman, J. M., Malaspina, D. & Susser, E. S.
(2002) Paternal age and risk of schizophrenia in adult offspring. American
Journal of Psychiatry 159:1528 – 33. [aMCK]
Brown, A. S. & Susser, E. S. (2002) In utero infection and adult schizophrenia.
Mental Retardation and Developmental Disabilities Research Review 8:51 –57.
[R-PB, rMCK]
Brown, C. & Braithwaite, V. A. (2004) Effects of predation pressure on the cognitive
ability of the poeciliid Brachyraphis episcopi. Behavioural Ecology 16(2):482 –
87. [EIK]
Brown, G. W., Birley, J. L. & Wing, J. K. (1972) Influence of family life on the
course of schizophrenic disorders: A replication. British Journal of Psychiatry
121:241 – 58. [R-PB]
Brown, J. S. (2003) Identification of candidate genes for schizophrenia based on
natural resistance to infectious diseases. Acta Neuropsychiatrica 15:108 – 14.
Brown, S. M., Peet, E., Manuck, S. B., Williamson, D. E., Dahl, R. E., Ferrell, R. E.
& Hariri, A. R. (2005) A regulatory variant of the human tryptophan
hydroxylase-2 gene biases amygdala reactivity. Molecular Psychiatry
10(9):884 – 88. [DCA]
Brugger, P. (2001) From haunted brain to haunted science: A cognitive neuroscience view of paranormal and pseudoscientific thought. In: Hauntings and
poltergeists: Multidisciplinary perspectives, ed. J. Houran & R. Lange,
pp. 195 – 213. McFarland. [BC]
Brüne, M. (2002) Toward an integration of interpersonal and biological processes:
Evolutionary psychiatry as an empirically testable framework for psychiatric
research. Psychiatry 65:48 – 57. [MB]
(2003) Erotomania revisited: Clues of its origin from evolutionary theory. In:
Advances in Psychology Research, vol. 21, ed. S. P. Shohov. Nova Science.
(2004) Schizophrenia: An evolutionary enigma? Neuroscience and Biobehavioral
Review 28:41– 53. [aMCK]
Buck, C., Hobbs, G. E., Simpson, H. & Waklin, J. M. (1975) Fertility of the sibs of
schizophrenic patients. British Journal of Psychiatry 127:235– 39. [rMCK]
Buist-Bouwman, M. A., de Graaf, R., Vollebergh, W. A. M. & Ormel, J. (2005)
Comorbidity of physical and mental disorders and the effect on work-loss days.
Acta Psychiatrica Scandinavica 111:436– 43. [aMCK]
Bulayeva, K. B., Leal, S. M., Pavlova, T. A., Kurbanov, R. M., Glatt, S. J., Bulayev,
O. A. & Tsuang, M. T. (2005) Mapping genes of complex psychiatric diseases
in Daghestan genetic isolates. American Journal of Medical Genetics. Part B:
Neuropsychiatric Genetics 132:76 – 84. [aMCK]
Bürger, R. (2000) The mathematical theory of selection, recombination, and
mutation. Wiley. [aMCK]
Bürger, R., Wagner, G. P. & Stettinger, F. (1989) How much heritable variation can
be maintained in finite populations by mutation-selection balance? Evolution
43:1748– 66. [OM]
Burns, J. K. (2004) An evolutionary theory of schizophrenia: Cortical connectivity,
metarepresentation, and the social brain. Behavioral and Brain Sciences
27(6):831 – 55; Discussion, 855 – 85. [BC, arMCK, JJM]
Burr, W. A., Falek, A., Strauss, L. T. & Brown, S. B. (1979) Fertility in psychiatric
outpatients. Hospital and Community Psychiatry 30:527– 31. [aMCK]
Burt, A. & Trivers, R. (2000) Genetic conflicts in genomic imprinting. Proceedings
of the Royal Society of London B: Biological Sciences 265:141 – 46. [SWG]
(2006) Genes in conflict: The biology of selfish genetic elements. Harvard
University Press. [BC]
Buss, D. M. (1984) Marital assortment for personality dispositions: Assessment with
three data sources. Behavior Genetics 14:111 – 23. [DMB]
(1991) Conflict in married couples: Personality predictors of anger and upset.
Journal of Personality 59:663 – 88. [DMB]
(1995) Evolutionary psychology: A new paradigm for psychological science.
Psychological Inquiry 6:1– 30. [NBA, aMCK]
(1996) Social adaptation and five major factors of personality. In: The five-factor
model of personality: Theoretical perspectives, ed. J. S. Wiggins, pp. 180 – 207.
Guilford Press. [DMB]
(2000) The dangerous passion: Why jealousy is necessary in love and sex. Free
Press. [JAE, rMCK]
(2003) The evolution of desire: Strategies of human mating (revised edition).
Basic Books. [DMB]
Buss, D. M., Abbott, M., Angleitner, A., Biaggio, A., Blanco-Villasenor, A.,
Bruchon-Schweitzer, M. [& 45 additional authors] (1990) International
preferences in selecting mates: A study of 37 societies. Journal of Cross
Cultural Psychology 21:5 – 47. [DMB]
Buss, D. M. & Barnes, M. L. (1986) Preferences in human mate selection. Journal
of Personality and Social Psychology 50:559– 70. [DMB]
Buss, D. M. & Greiling, H. (1999) Adaptive individual differences. Journal of
Personality 67:209 – 43. [DMB]
Campbell, D. T. (1974) Evolutionary epistemology. In: The philosophy of Karl
Popper, ed. P. A. Schilpp. Open Court. [JW]
Campbell, T. A. (2005) The cognitive neuroscience of auditory distraction. Trends
in Cognitive Sciences 9:3– 5. [TC]
Camperio-Ciani, A., Corna, F. & Capiluppi, C. (2004) Evidence for maternally
inherited factors favouring male homosexuality and promoting female
fecundity. Proceeding of the Royal Society of London B 271:2217– 21. [EIK]
Cannon, T. D. & Keller, M. C. (2005) Endophenotypes in genetic analyses of
mental disorders. Annual Review of Clinical Psychology 2:267 – 90. [aMCK]
Cardinal, R. N. (2004) Waiting for better things. The Psychologist 17(12):684 –87.
Cardno, C. & Murray, R. (2003) The “classical” genetic epidemiology of schizophrenia. In: The epidemiology of schizophrenia, ed. R. M. Murray, P. B. Jones,
E. Susser, J. Van Os & M. Cannon. Cambridge University Press. [JP]
Cargill, M., Altshuler, D., Ireland, J., Sklar, P., Ardlie, K., Patil, N., Lane, C. R.,
Lim, E. P., Kalyanaraman, N., Nemesh, J., Ziaugra, L., Friedland, L., Rolfe, A.,
Warrington, J., Lipshutz, R., Daley, G. Q. & Lander, E. S. (1999) Characterization of single-nucleotide polymorphisms in coding regions of human genes.
Nature Genetics 22:231 – 38. [aMCK]
Carroll, S. B. (2005) Endless forms most beautiful: The new science of evo devo and
the making of the animal kingdom. W. W. Norton. [JJM]
Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington, H.,
McClay, J., Mill, J., Martin, J., Braithwaite, A. & Poulton, R. (2003) Influence
of life stress on depression: Moderation by a polymorphism in the 5-HTT gene.
Science 301(5631):386 –89. [NBA, rMCK, JW]
Catts, V. S. & Catts, S. V. (2000) Apoptosis and schizophrenia: Is the tumour
suppressor gene, p53, a candidate susceptibility gene? Schizophrenia Research
41(3):405 – 15. [JJM]
Chakrabarti, S. & Fombonne, E. (2005) Pervasive developmental disorders in
preschool children: Confirmation of high prevalence. American Journal of
Psychiatry 162:1133 – 41. [MAG]
Chaleby, K. & Tuma, T. A. (1986) Cousin marriages and schizophrenia in Saudi
Arabia. British Journal of Psychiatry 150:547 – 49. [aMCK]
Charlesworth, B. (1984) The evolutionary genetics of life histories. In: Evolutionary
ecology, ed. B. Shorrocks, pp. 117 – 33. Blackwell. [aMCK]
(1987) The heritability of fitness. In: Sexual selection: Testing the alternatives, ed.
J. W. Bradbury & M. B. Andersson, pp. 21 – 40. Wiley. [aMCK]
(1990) Optimization models, quantitative genetics, and mutation. Evolution
44:520 – 38. [aMCK]
Charlesworth, B. & Hughes, K. A. (1999) The maintenance of genetic variation in
life-history traits. In: Evolutionary genetics: From molecules to morphology,
vol. 1, ed. R. S. Singh & C. B. Krimbas, pp. 369 – 92. Cambridge University
Press. [aMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Cheng, R., Jou, S. H., Loth, J. E., Nee, J., Iossifov, I., Blumenthal, R., Sharpe, L.,
Kanyas, K., Lerer, B., Lilliston, B., Smith, M., Trautman, K., Gilliam, T. C.,
Endicott, J. & Baron, M. (2006) Genome-wide linkage scan in a large bipolar
disorder sample from the National Institute of Mental Health genetics
initiative suggests putative loci for bipolar disorder, psychosis, suicide, and
panic disorder. Molecular Psychiatry 11:252 – 60. [MV]
Cho, H. J., Meira-Lima, I., Cordeiro, Q., Michelon, L., Sham, P. C., Vallada, H. &
Collier, D. A. (2005) Population-based and family-based studies on the serotonin transporter gene polymorphisms and bipolar disorder: A systematic
review and meta-analysis. Molecular Psychiatry 10:771 – 81. [rMCK]
Chodoff, P. (2002) The medicalization of the human condition. Psychiatric Services
53:627 – 28. [AT]
Chuang, J. H. & Li, H. (2004) Functional bias and spatial organization of genes in
mutational hot and cold regions in the human genome. PLoS Biology 2(2):E29.
Clark, L. A. (2005) Temperament as a unifying basis for personality and
psychopathology. Journal of Abnormal Psychology 114(4):505 – 21.
Cliff, D. & Miller, G. F. (2006) Visualizing coevolution with CIAO plots. Artificial
Life 12:199 – 202. [rMCK]
Cochran, G. M., Ewald, P. W. & Cochran, K. D. (2000) Infectious causation of
disease: An evolutionary perspective. Perspectives in Biology and Medicine
43:406 – 48. [rMCK]
Cochran, G. M., Hardy, J. & Harpending, H. (in press) Natural history of Ashkenazi
intelligence. Journal of Biological Science. [rMCK]
Cohen, D., Pichard, N., Tordjman, S., Baumann, C., Burglen, L., Excoffier, E.,
Lazar, G., Mazet, P., Pinquier, C., Verloes, A. & Héron, D. (2005) Specific
genetic disorders and autism: Clinical contribution towards their identification.
Journal of Autism and Developmental Disorders 35:103 – 16. [BC]
Cohen, H. (1981) The evolution of the concept of disease. In: Concepts of health
and disease: Interdisciplinary perspectives, ed. A. L. Caplan, H. T. Engelhardt,
Jr. & J. J. McCartney, pp. 209 – 20. Addison-Wesley. [AT]
Cohen, J. C., Kiss, R. S., Pertsemlidis, A., Marcel, Y. L., McPherson, R. &
Hobbs, H. H. (2004) Multiple rare alleles contribute to low plasma levels of
HDL cholesterol. Science 305(5685):869– 72. [DCA]
Cohen, J. C., Pertsemlidis, A., Fahmi, S., Esmail, S., Vega, G. L., Grundy, S. M. &
Hobbs, H. H. (2006) Multiple rare variants in NPC1L1 associated with
reduced sterol absorption and plasma low-density lipoprotein levels.
Proceedings of the National Academy of Sciences USA 103(6):1810– 15. [DCA]
Corbo, R. M. & Scacchi, R. (1999) Apolipoprotein E (APOE) allele distribution in
the world: Is APOE 4 a “thrifty” allele? Annals of Human Genetics 63:301 – 10.
Corradi, J. P., Ravyn, V., Robbins, A. K., Hagan, K. W., Peters, M. F., Bostwick, R.,
Buono, R. J., Berrettini, W. H. & Furlong, S. T. (2005) Alternative transcripts
and evidence of imprinting of GNAL on 18p11.2. Molecular Psychiatry
10:1017 – 25. [SWG]
Cosmides, L. & Tooby, J. (1987) From evolution to behavior. In: The latest on the
best: Essays on evolution and optimality, ed. J. Dupre. MIT Press. [JW]
(n.d.) Evolutionary psychology: A primer. Available at: http://www.psych.ucsb.
edu/research/cep/primer.html. [aMCK]
Costas, J., Salas, A., Phillips, C. & Carracedo, A. (2005) Human genome-wide
screen of haplotype-like blocks of reduced diversity. Gene 349:219 – 25. [BC]
Coventry, W. L. & Keller, M. C. (2005) Estimating the extent of parameter bias in
the classical twin design: A comparison of parameter estimates from extended
twin-family and classical twin designs. Twin Research and Human Genetics
8:214 –23. [aMCK]
Crabbe, J. C. (1999) Animal models in neurobehavioral genetics: Methods for
estimating genetic correlation. In: Neurobehavioral genetics: Methods
and applications, ed. B. C. Jones & P. Mormede, pp. 121 – 38.
CRC Press. [DCA]
Craddock, N. & Jones, I. (1999) Genetics of bipolar disorder. Journal of Medical
Genetics 36:585 – 94. [rMCK]
Craddock, N., O’Donovan, M. C. & Owen, M. J. (2005) The genetics of
schizophrenia and bipolar disorder: Dissecting psychosis. Journal of Medical
Genetics 42:193 – 204. [aMCK]
Crow, J. F. (2000) The origins, patterns, and implications of human spontaneous
mutation. Nature Reviews Genetics 1:40 – 47. [aMCK]
Crow, J. F. & Kimura, M. (1970) An introduction to population genetics theory.
Harper & Row. [aMCK]
Crow, T. J. (1995a) A continuum of psychosis, one human gene, and not much
else – the case for homogeneity. Schizophrenia Research 17(2):135– 45.
Available at:¼pubmed&
itool¼pubmed_docsum [DRW]
(1995b) A theory of the evolutionary origins of psychosis. European Neuropsychopharmacology 5(Suppl.):59 – 63. [JW]
(1997) Is schizophrenia the price that Homo sapiens pays for language?
Schizophrenia Research 28:127 – 41. [BC]
(2000) Schizophrenia as the price that Homo sapiens pays for language: A
resolution to the central paradox in the origin of the species. Brain Research:
Brain Research Reviews 31(2 – 3):118 – 29. [TC, arMCK, JJM]
Crowley, P. H. (2003) Origins of behavioural variability: Categorical and discriminative assessments in serial contests. Animal Behaviour 66:427 – 40. [JSP]
Curtisinger, J. W., Service, P. M. & Prout, T. (1994) Antagonistic pleiotropy,
reversal of dominance, and genetic polymorphism. American Naturalist
144:210 –28. [arMCK]
Dale, R. C. & Heyman, I. (2002) Post-streptococcal autoimmune psychiatric and
movement disorders in children. British Journal of Psychiatry 181:188 – 90.
Daly, G., Hawi, Z., Fitzgerald, M. & Gill, M. (1999) Mapping susceptibility loci in
attention deficit hyperactivity disorder: Preferential transmission of parental
alleles at DAT1, DBH and DRD5 to affected children. Molecular Psychiatry
4(2):192 – 96. [JW]
Daly, M. & Wilson, M. (1983) Sex, evolution, and behavior. Willard Grant. [aMCK]
(1988) Homicide. Aldine de Gruyter. [JAE]
David, A. S. & Halligan, P. W. (2000) Cognitive neuropsychiatry: Potential for
progress. Journal of Neuropsychiatry and Clinical Neurosciences 12(4):506–
10. [JW]
Dayan, P. & Williams, J. (2006) Putting the computation back into computational
modelling. Pharmacopsychiatry 39(Suppl. 1):S50– S51. [JW]
de Catanzaro, D. (1980) Human suicide: A biological perspective. Behavioral and
Brain Sciences 3:265 – 90. [MV]
(1981) Suicide and self-damaging behavior: A sociobiological perspective.
Academic Press. [MV]
DeLisi, L. E., Shaw, S. H., Crow, T. J., Shields, G., Smith, A. B., Larach, V. W.,
Wellman, N., Loftus, J., Nanthakumar, B., Razi, K., Stewart, J., Comazzi, M.,
Vita, A., Heffner, T. & Sherrington, R. (2002) A genome-wide scan for linkage
to chromosomal regions in 382 sibling pairs with schizophrenia or schizoaffective disorder. American Journal of Psychiatry 159:803– 12. [SWG]
Depue, R. A. & Collins, P. F. (1999) Neurobiology of the structure of personality:
Dopamine, facilitation of incentive motivation, and extraversion. Behavioral
and Brain Sciences 22:491 – 569. [NBA]
Diller, K. C., Gilbert, W. A. & Kocher, T. D. (2002) Selective sweeps in the human
genome: A starting point for identifying genetic differences between modern
humans and chimpanzees. Molecular Biology and Evolution 19:2342– 45. [BC]
Ding, Y. C., Chi, H. C., Grady, D. L., Morishima, A., Kidd, J. R., Kidd, K. K.,
Flodman, P., Spence, M. A., Schuck, S., Swanson, J. M., Zhang, Y. P. & Moyzis,
R. K. (2002) Evidence of positive selection acting at the human dopamine
receptor D4 gene locus. Proceedings of the National Academy of Sciences USA
99(1):309 – 14. [BC, JW]
Dingemanse, N. J., Bopth, C., Drent, P. J. & Tinbergen, J. M. (2004) Fitness
consequences of avian personalities in a fluctuating environment. Proceedings
of the Royal Society of London, B: Biological Sciences 271:847 –52. [NBA]
Dragovic, M. & Hammond, G. (2005) Handedness in schizophrenia: A quantitative
review of evidence. Acta Psychiatrica Scandinavica 111:410– 19. [AP]
Draguns, J. G. (1982) Methodology in cross-cultural psychopathology. In: Culture
and psychopathology, ed. I. Al-Issa. University Park Press. [rMCK]
Dubrovsky, B. (2002) Evolutionary psychiatry: Adaptationist and nonadaptationist
conceptualizations. Progress in Neuro-Psychopharmacology and Biological
Psychiatry 26:1 – 19. [MB, AT]
Eaves, L. J., Heath, A. C., Martin, N. G., Maes, H. H., Neale, M. C., Kendler, K. S.,
Kirk, K. & Corey, L. (1999) Comparing the biological and cultural inheritance
of personality and social attitudes in the Virginia 30,000 study of twins and their
relatives. Twin Research 2:62 – 80. [aMCK]
Eaves, L. J., Heath, A. C., Neale, M. C., Hewitt, J. K. & Martin, N. G. (1998) Sex
differences and non-additivity in the effects of genes on personality. Twin
Research 1:131 – 37. [DMB]
Eaves, L. J., Last, K. A., Young, P. A., & Martin, N. G. (1978) Model-fitting
approaches to the analysis of human behavior. Heredity 41:249 – 320. [aMCK]
Eaves, L. J., Martin, N. G., Heath, A. C., Hewitt, J. K. & Neale, M. C. (1990)
Personality and reproductive fitness. Behavior Genetics 20:563 – 68. [NBA,
Egan, M. F., Goldberg, T. E., Kolachana, B. S., Callicott, J. H., Mazzanti, C. M.,
Straub, R. E., Goldman, D. & Weinberger, D. R. (2001) Effect of COMT
Val(108/158) Met genotype on frontal lobe function and risk for schizophrenia.
Proceedings of the National Academy of Sciences USA 98:6917– 22. [TC]
Eiben, O. G., Kelemen, A., Pethö, B. & Felsövályi, Á. (1980) Physique of
endogenous psychotic female patients. Anthropologiai Közlemények 24:77– 82.
Eley, T. C. & Stevenson, J. (2000) Specific life events and chronic experiences
differentially associated with depression and anxiety in young twins. Journal of
Abnormal Child Psychology 28(4):383 – 94. [JW]
Elman, J. L. (2005) Connectionist models of cognitive development: Where next?
Trends in Cognitive Sciences 9(3):111 – 17. [JW]
Embry, D. D. (2002) Nurturing the genius of genes: The new frontier of education,
therapy, and understanding of the brain. Brain and Mind 3:101– 32. [aMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Enard, W., Przeworski, M., Fisher, S. E., Lai, C. S., Wiebe, V., Kitano, T., Monaco,
A. P. & Pääbo, S. (2002) Molecular evolution of FOXP2, a gene involved in
speech and language. Nature 418:869– 72. [BC]
Endler, J. A. (1986) Natural selection in the wild. Princeton University Press.
Enoch, M. D. & Trethowan, W. H. (1979). The Othello syndrome. In: Uncommon
psychiatric syndromes, 2nd edition. John Wright. [JAE]
Erlenmeyer-Kimling, L., Nicol, S., Rainer, J. D. & Deming, E. (1969) Changes in
fertility rates of schizophrenic patients in New York. American Journal of
Psychiatry 125:916 –27. [aMCK]
Evans, P. D., Gilbert, S. L., Mekel-Bobrov, N., Vallender, E. J., Anderson, J. R.,
Vaez-Azizi, L. M., Tishkoff, S. A., Hudson, R. R. & Lahn, B. T. (2005)
Microcephalin, a gene regulating brain size, continues to evolve adaptively in
humans. Science 309:1717 – 20. [rMCK]
Evenden, J. L. (1999) Varieties of impulsivity. Psychopharmacology 146:348 – 61.
Ewens, W. J. (2001) Fundamental theorem of natural selection. In: Encyclopedia of
genetics, ed. S. Brenner & J. H. Miller, pp. 733 – 38. Academic Press. [OM]
Eyre-Walker, A. & Keightley, P. D. (1999) High genomic deleterious mutation rates
in hominids. Nature 397:344 – 47. [aMCK]
Eysenck, H. J. (1947) Dimensions of personality. Kegan Paul. [JAS]
(1953) The structure of human personality. Methuen. [JAS]
Falconer, D. S. & Mackay, T. F. C. (1996) Introduction to quantitative genetics, 4th
edition. Addison Wesley Longman. [arMCK]
Fananás, L. & Bertranpetit, J. (1995) Reproductive rates in families of schizophrenic patients in a case-control study. Acta Psychiatrica Scandinavica
91:202 – 204. [BC, arMCK]
Faraone, S. V., Doyle, A. E., Mick, E. & Biederman, J. (2001) Meta-analysis of the
association between the 7-repeat allele of the dopamine D(4) receptor gene
and attention deficit hyperactivity disorder. American Journal of Psychiatry
158(7):1052 – 57. [JW]
Faraone, S. V., Perlis, R. H., Doyle, A. E., Smoller, J. W., Goralnick, J. J., Holmgren,
M. A. & Sklar, P. (2005) Molecular genetics of attention-deficit/hyperactivity
disorder. Biological Psychiatry 57(11):1313– 23. [JW]
Farmer, A., McGuffin, P. & Williams, J. (2002) Measuring psychopathology. Oxford
University Press. [aMCK]
Farrer, L. A., Cupples, L. A., Kiely, D. K., Conneally, P. M. & Myers, R. H. (1992)
Inverse relationship between age at onset of Huntington disease and paternal
age suggests involvement of genetic imprinting. American Journal of Human
Genetics 50:528 –35. [BC, rMCK]
Faurie, C. & Raymond, M. (2005) Handedness, homicide and negative frequencydependent selection. Proceedings of the Royal Society of London, B: Biological
Sciences 272:25 – 28. Available at
15875566&query_hl¼4&itool ¼ pubmed_docsum. [DRW]
Fay, J. C., Wyckoff, G. J. & Wu, C. (2001) Positive and negative selection on the
human genome. Genetics 158:1227– 34. [aMCK]
Fehr, E. & Fischbacher, U. (2003) The nature of human altruism. Nature 425:785 –
91. [AP]
Finch, C. E. & Sapolsky, R. M. (1999) The evolution of Alzheimer disease, the
reproductive schedule, and APOE isoforms. Neurobiology of Aging 20:407 –
28. [aMCK]
Fisher, J. E., Mohanty, A., Herrington, J. D., Koven, N. S., Miller, G. A. & Heller, W.
(2004) Neuropsychological evidence for dimensional schizotypy: Implications
for creativity and psychopathology. Journal of Research in Personality
38:24– 31. [BC]
Fisher, R. A. (1930/1999) The genetical theory of natural selection. Clarendon
Press. (First edition: Clarendon Press,1930; Reprint: (1999) The genetical
theory of natural selection: A complete variorum edition, ed. J.H. Bennett.
Oxford University Press.] [aMCK, OM]
Folley, B. S., Doop, M. L. & Park, S. (2003) Psychoses and creativity: Is the missing
link a biological mechanism related to phospholipids turnover? Prostaglandins,
Leukotrienes, and Essential Fatty Acids 69:467 – 76. [BC]
Folley, B. S. & Park, S. (2005) Verbal creativity and schizotypal personality in
relation to prefrontal hemispheric laterality: A behavioral and near-infrared
optical imaging study. Schizophrenia Research 80:271 – 82. [BC]
Folstein, S. E. & Rosen-Sheidley, B. (2001) Genetics of autism: Complex aetiology
for a heterogeneous disorder. Nature Reviews Genetics 2:943 – 55. [aMCK]
Foucault, M. (1965/1988) Madness and civilization. Random House. [JP]
Francks, C., DeLisi, L. E., Shaw, S. H., Fisher, S. E., Richardson, A. J., Stein, J. F.
& Monaco, A. P. (2003) Parent-of-origin effects on handedness and schizophrenia susceptibility on chromosome 2p12-q11. Human Molecular Genetics
12:3225 – 30. [BC, SWG]
Friedman, H. S., Tucker, J. S., Schwartz, J. E., Tomlinson-Keasey, C., Martin, L. R.,
Wingard, D. L. & Criqui, M. H. (1995) Psychosocial and behavioral predictors
of longevity: The aging and death of the “termites.” American Psychologist
50:69– 78. [DMB]
Frith, U. (1989) Autism: Explaining the enigma. Blackwell. [MAG]
Gadanho, S. C. (2003) Learning behavior-selection by emotions and cognition in a
multi-goal robot task. Journal of Machine Learning Research 4:385– 412. [JW]
Gangestad, S. W. (1997) Evolutionary psychology and genetic variation: Nonadaptive, fitness-related, adaptive. In: Characterizing human psychological
adaptations, ed. G. R. Bock & G. Cardew, pp. 212 – 23. Wiley. [SWG]
(2003) Sexually antagonistic coevolution: Theory, evidence, and implications for
human patterns of mating and fertility. In: Offspring: The biodemography of
fertility-related behavior, ed. K. D. Wachter & R. Bulatao, pp. 224 – 59.
National Academy of the Sciences. [rMCK]
Gangestad, S. W. & Yeo, R. A. (1997) Behavioral genetic variation, adaptation and
maladaptation: An evolutionary perspective. Trends in Cognitive Science
1:103– 108. [arMCK, SWG]
Garcı́a-Dorado, A., Caballero, A. & Crow, J. F. (2003) On the persistence and
pervasiveness of a new mutation. Evolution 57:2644– 46. [aMCK]
Garcı́a-Dorado, A., López-Fanjul, C. & Caballero, A. (2004) Rates and effects of
deleterious mutations and their evolutionary consequences. In: Evolution:
From molecules to ecosystems, ed. A. Moya & E. Font. Oxford University Press.
Gardner, M., Gonzalez-Neira, A., Lao, O., Calafell, F., Bertranpetit, J. & Comas, D.
(2006) Extreme population differences across Neuregulin 1 gene, with implications for association studies. Molecular Psychiatry 11:66– 75. [BC]
Gardner, R., Jr. & Price, J. S. (1999) Sociophysiology and depression. In: The
Interactional Nature of Depression: Advances in Interpersonal Approaches, ed.
T. Joiner & J. C. Coyne. APA. [DRW]
Garrigan, D. & Hedrick, P. W. (2003) Detecting adaptive molecular polymorphism:
Lessons from the MHC. Evolution 57:1707 – 22. [aMCK]
Geist, V. (1989) Environmentally guided phenotype plasticity in mammals and
some of its consequences to theoretical and applied biology. In: Alternative
life-history styles of animals, ed. M. N. Bruton. Kluwer Academic. [JSP]
Gernsbacher, M. A., Dawson, M. & Goldsmith, H. H. (2005) Three reasons not to
believe in an autism epidemic. Current Directions in Psychological Science
14:55– 58. [MAG]
Gilbert, P. (1992) Depression: The evolution of powerlessness. Guilford Press.
(1998) Evolutionary psychopathology: Why isn’t the mind designed better than it
is? British Journal of Medical Psychiatry 71:353– 73. [aMCK]
Gindilis, V. M., Gainullin, R. G. & Shmaonova, L. M. (1989) Genetico-demographic
patterns of the prevalence of various forms of endogenous psychoses. Genetika
25:734 – 43. [aMCK]
Gingerich, K. J., Turnock, P., Litfin, J. K. & Rosen, L. A. (1998) Diversity and
attention deficit hyperactivity disorder. Journal of Clinical Psychology
54(4):415 – 26. [EIK]
Glantz, L. A., Gilmore, J. H., Lieberman, J. A. & Jarskog, L. F. (2006) Apoptotic
mechanisms and the synaptic pathology of schizophrenia. Schizophrenia
Research 81(1):47 – 63. [JJM]
Glatt, S. J., Faraone, S. V. & Tsuang, M. T. (2003) Schizophrenia is not associated
with DRD4 48-base-pair-repeat length or individual alleles: Results of a metaanalysis. Biological Psychiatry 15:629 – 35. [rMCK]
Glenn, S. & Cunningham, C. (2001) Evaluation of self by young people with Down
syndrome. International Journal of Disability, Development and Education
48:163 – 77. [rMCK]
Golden, R. N., Gaynes, B. N., Ekstrom, R. D., Hamer, R. M., Jacobsen, F. M.,
Suppes, T., Wisner, K. L. & Nemeroff, C. B. (2005) The efficacy of light
therapy in the treatment of mood disorders: A review and meta-analysis of the
evidence. American Journal of Psychiatry 162:656– 62. [JAS]
Goodwin, F. K. & Jamison, K. R. (1990) Manic-depressive illness. Oxford University
Press. [JAS]
Gottesman, I. I. (1991) Schizophrenia genesis. W. H. Freeman. [rMCK]
Gottesman, I. I. & Erlenmeyer-Kimling, L. (2001) Family and twin strategies
as a head start in defining prodromes and endophenotypes for
hypothetical early-interventions in schizophrenia. Schizophrenia Research
51:93– 102. [AP]
Gottesman, I. I. & Gould, T. D. (2003) The endophenotype concept in psychiatry:
Etymology and strategic intentions. American Journal of Psychiatry 160:636–
45. [aMCK]
Gottesman, I. I. & Shields, J. A. (1967) A polygenic theory of schizophrenia. Proceedings of the National Academy of Sciences USA 58:199– 205. [aMCK]
Gregory, W. K. (1928) The upright posture of man: A review of its origin and
evolution. Proceedings of the American Philosophical Society 67(4):339– 77.
Guisinger, S. (2003) Adapted to flee famine: Adding an evolutionary perspective on
Anorexia Nervosa. Psychological Review 110:745 – 61. [aMCK, AP]
Hagen, E. H. (1999) The function of postpartum depression. Evolution and Human
Behavior 20:325 –59. [rMCK]
Haldane, J. B. S. (1949) Disease and evolution. Ricerca Scientifica 19(Suppl. A):3 –
11. [aMCK]
Haig, D. (1993) Genetic conflicts in human pregnancy. Quarterly Review of Biology
68:495 – 532. [SWG]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
(2000) The kinship theory of genomic imprinting. Annual Review of Ecological
Systems 31:9 – 32. [SWG]
Hall, B. K. (1999) Evolutionary developmental biology, 2nd edition. Kluwer. [NBA]
Halushka, M. K., Tan, J. B., Bentley, K., Hsie, L., Shen, N. P., Weder, A.,
Cooper, R., Lipshutz, R. & Chakravarti, A. (1999) Patterns of single-nucleotide
polymorphism at candidate genes for blood-pressure homeostasis. Nature
Genetics 22:239 – 47. [aMCK]
Hamblin, M. T., Thompson, E. E. & DiRienzo, A. (2002) Complex signatures of
natural selection at the Duffy blood group locus. American Journal of Human
Genetics 70:369 – 83. [aMCK]
Hamilton, W. D. (1964) The genetical evolution of social behaviour. I and II.
Journal of Theoretical Biology 7:1– 52. [AP]
Hardcastle, V. G. (2004) Schizophrenia: A benign trait. Behavioral and Brain
Sciences 27:859 – 60. [aMCK]
Hare, E. H. & Moran, P. A. (1979) Raised parental age in psychiatric patients:
Evidence for the constitutional hypothesis. British Journal of Psychiatry
134:169 – 77. [aMCK]
Harrison, P. J. & Weinberger, D. R. (2005) Schizophrenia genes, gene expression,
and neuropathology: On the matter of their convergence. Molecular Psychiatry
10:40 –68. [BC]
Haselton, M. & Miller, G. F. (2006) Women’s fertility across the cycle increases the
short-term attractiveness of creative intelligence compared to wealth. Human
Nature 17:50 – 73. [aMCK]
Haslam, N. (2003) Categorical versus dimensional models of mental disorder: The
taxometric evidence. Australian and New Zealand Journal of Psychiatry
37:696 – 704. [aMCK]
Haukka, J., Suvisaari, J. & Lönnqvist, J. (2003) Fertility of patients with
schizophrenia, their siblings, and the general population: A cohort study
from 1950 to 1959 in Finland. American Journal of Psychiatry 160:460– 63.
[BC, arMCK]
Haverkamp, F., Propping, P. & Hilger, T. (1982) Is there an increase of reproduction rates in schizophrenics? I. Critical review of the literature. European
Archives of Psychiatry and Clinical Neuroscience 232(5):439 – 50. [rMCK]
Hawi, Z., Segurado, R., Conroy, J., Sheehan, K., Lowe, N., Kirley, A., Shields, D.,
Fitzgerald, M., Gallagher, L. & Gill, M. (2005) Preferential transmission of
paternal alleles at risk genes in attention-deficit/hyperactivity disorder.
American Journal of Human Genetics 77:958 – 65. [SWG]
Hedrick, P. W. (1999) Antagonistic pleiotropy and genetic polymorphism: A perspective. Heredity 82:126 – 32. [arMCK]
Hermelin, B. & O’Connor, N. (1970) Psychological experiments with autistic children. Pergamon. [MAG]
Hibbeln, J. R. (1998) Fish consumption and major depression. Lancet 351:1213.
Hinton, G. E. (1987) How learning can guide evolution. Complex Systems 1:495 –
502. [JW]
Hoffman, R. E., Hampson, M., Varanko, M. & McGlashan, T. H. (2004) Auditory
hallucinations, network connectivity, and schizophrenia. Behavioral and Brain
Sciences 27(6):860 –61. [BC]
Hollister, J. M., Laing, P. & Mednick, S. A. (1996) Rhesus incompatibility as a risk
factor for schizophrenia in male adults. Archives of General Psychiatry 53:19 –
24. [AP]
Horrobin, D. (2002) The madness of Adam and Eve: How schizophrenia shaped
humanity. Bantam. [arMCK]
Horrobin, D. F. (1998) Schizophrenia: The illness that made us human. Medical
Hypotheses 50:269 – 88. [BC]
Horwitz, A. V. & Wakefield, J. C. (2005) The age of depression. Public Interest
158:39 – 58. [JCW]
(2006) The epidemic in mental illness: Clinical fact or survey artifact? Contexts:
Understanding People in Their Social Worlds 5:19 – 23. [JCW]
Houle, D. (1992) Comparing evolvability and variability of quantitative traits.
Genetics 130:195– 205. [aMCK]
(1998) How should we explain variation in the genetic variance of traits? Genetica
102:241 – 53. [aMCK]
Hovatta, I., Varilo, T., Suvisaari, J., Terwilliger, J. D., Ollikainen, V., Arajärvi, R.,
Juvonen, H., Kokko-Sahin, M. L., Vaisanen, L., Mannila, H., Lonnqvist, J. &
Peltonen, L. (1999) A genomewide screen for schizophrenia genes in an isolated Finnish subpopulation, suggesting multiple susceptibility loci. American
Journal of Human Genetics 65:1114– 24. [TC]
Howard, L. M., Kumar, C., Leese, M. & Thornicroft, G. (2002) The general fertility
rate in women with psychotic disorders. American Journal of Psychiatry
159:991 – 97. [aMCK]
Hughes, K. A. & Burleson, M. H. (2000) Evolutionary causes of genetic variation in
fertility and other fitness components. In: Genetic influences on human sexuality and fertility, ed. J. L. Rodgers, D. C. Rowe & W. Miller, pp. 7 –34. Kluwer
Academic. [aMCK]
Hull, D. L., Langman, R. E. & Glenn, S. S. (2001) A general account of selection:
Biology, immunology, and behavior. Behavioral and Brain Sciences
24:511 – 73. [rMCK]
Huxley, J., Mayr, E., Osmond, H. & Hoffer, A. (1964) Schizophrenia as a genetic
morphism. Nature 204:220– 21. [aMCK]
Hyttinen, V., Kaprio, J., Kinnunen, L., Koskenvuo, M. & Tuomilehto, J. (2003)
Genetic liability of type 1 diabetes and the onset age among 22,650 young
Finnish twin pairs: A nationwide follow-up study. Diabetes 52:1052 –55. [OM]
Inoue, K. & Lupski, J. R. (2003) Genetics and genomics of behavioral and psychiatric disorders. Current Opinion in Genetics and Development 13:303 – 309.
Jablensky, A., Sartorius, N., Ernberg, G., Anker, M., Korten, A., Cooper, J. E.,
Day, R. & Bertelsen, A. (1992) Schizophrenia: Manifestations, incidence and
course in different cultures. A World Health Organization ten-country study.
Psychological Medicine Monographs 20:1 – 97. [aMCK, AP]
Jacobi, F., Wittchen, H. U., Hölting, C., Höfler, M., Pfister, H., Müller, N. &
Lieb, R. (2004) Prevalence, co-morbidity and correlates of mental disorders in
the general population: Results from the German Health Interview and
Examination Survey (GHS). Psychological Medicine 34:597 – 611. [aMCK]
Jamison, K. R. (1993) Touched with fire: Manic depressive illness and the artistic
temperament. Simon & Schuster. [aMCK]
(2005) Exuberance: The passion for life. Knopf. [JAS]
Jansson, M., McCarthy, S., Sullivan, P. F., Dickman, P., Andersson, B., Oreland, L.,
Schalling, M. & Pedersen, N. L. (2005) MAOA haplotypes associated with
thrombocyte-MAO activity. BMC Genetics 6:46. [BC]
Jarvik, L. F. & Deckard, B. S. (1977) The Odyssean personality. A survival advantage for carriers of genes predisposing to schizophrenia? Neuropsychobiology
3:179– 91. [aMCK]
Jaspers, K. (1923) Allgemeine Psychopathologie, 3rd edition. Springer. [aMCK]
Jenkins, C. & Miller, G. F. (2005) Top 10 reasons why orgasm should be classed as a
basic human emotion. Paper presented at the annual meeting of the Human
Behavior and Evolution Society, Austin, TX, June 14, 2005. [rMCK]
Jensen, P. S., Mrazek, D., Knapp, P. K., Steinberg, L., Pfeffer, C., Schowalter, J. &
Shapiro, T. (1997) Evolution and revolution in child psychiatry: ADHD as a
disorder of adaptation. Journal of the American Academy of Child and
Adolescent Psychiatry 36(12):1672– 79. [JW]
Jeste, D. V., Del Carmen, R., Lohr, J. B. & Wyatt, R. J. (1985) Did schizophrenia
exist before the eighteenth century? Comprehensive Psychiatry 26:493 – 503.
Johansson, C., Jansson, M., Linner, L., Yuan, Q., Pedersen, N. L., Blackwood, D.,
Barden, N., Kelsoe, J. & Schalling, M. (2001) Genetics of affective disorders.
European Neuropsychopharmacology 11:385– 94. [aMCK]
Jones, P. B., Murray, R. M. & Rodgers, B. (1995) Childhood risk factors for adult
schizophrenia in a general population birth cohort at age 43 years. In: Neural
development and schizophrenia, ed. S. A. Mednick & J. M. Hollister. Plenum
Press. [R-PB]
Jones, P. B., Rodgers, B., Murray, R. & Marmot, M. (1994) Child development risk
factors for adult schizophrenia in the British 1946 birth cohort. Lancet
344(8934):1398 – 402. [R-PB]
Just, M. A., Cherkassky, V. L., Keller, T. A. & Minshew, N. J. (2004) Cortical
activation and synchronization during sentence comprehension in
high-functioning autism: Evidence of underconnectivity. Brain 127:
1811 –21. [MAG]
Kaplan, H., Hill, K., Lancaster, J. & Hurtado, A. M. (2000) A theory of human life
history evolution: Diet, intelligence, and longevity. Evolutionary Anthropology: Issues, News, and Reviews 9:156– 85. [aMCK]
Karlsson, J. L. (1974) Inheritance of schizophrenia. Acta Psychiatrica Scandinavica
247(Suppl.):77– 88. [aMCK]
(1999) Relation of mathematical ability to psychosis in Iceland. Clinical Genetics
56:447 – 49. [AP]
Keightley, P. D. & Gaffney, D. J. (2003) Functional constraints and frequency of
deleterious mutations in noncoding DNA of rodents. Proceedings of the
National Academy of Sciences USA 100:13402 – 406. [aMCK]
Keightley, P. D. & Hill, W. G. (1987) Directional selection and variation in finite
populations. Genetics 117:573 – 82. [OM]
Keller, M. C. (2004) Evolutionary theories of schizophrenia must ultimately explain
the genes that predispose to it. Behavioral and Brain Sciences 27:861 – 62.
(in press) The role of mutation in human mating abilities and preferences. In:
Mating intelligence: Theoretical, experimental, and differential perspectives,
ed. G. Geher & G. Miller. Erlbaum. [comments on Burns (2004)] [aMCK]
Keller, M. C. & Coventry, W. L. (2005) Quantifying and addressing parameter
indeterminacy in the classical twin design. Twin Research and Human Genetics
8:201– 13. [aMCK]
Keller, M. C., Coventry, W. L., Heath, A. C. & Martin, N. G. (2005) Widespread
evidence for non-additive genetic variation in Cloninger’s and Eysenck’s personality dimensions using a twin plus sibling design. Behavioral Genetics
35:707 – 21. [DMB]
Keller, M. C. & Nesse, R. M. (2005) Is low mood an adaptation? Evidence for
subtypes with symptoms that match precipitants. Journal of Affective Disorders
86(1):27 – 35. [aMCK, JW]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
(2006) The evolutionary significance of depressive symptoms: Different adverse
situations lead to different depressive symptom patterns. Journal of Personality
and Social Psychology 91:316 – 30. [arMCK]
Kendler, K. S. (2005) “A gene for . . .”: The nature of gene action in psychiatric
disorders. American Journal of Psychiatry 162:1243 – 52. [aMCK]
Kendler, K. S. & Eaves, L. J. (1986) Models for the joint effect of genotype and
environment on liability to psychiatric illness. American Journal of Psychiatry
143:279 – 89. [NBA]
Kendler, K. S., Gardner, C. O. & Prescott, C. A. (2002) Toward a comprehensive
developmental model for major depression in women. American Journal of
Psychiatry 159(7):1133 – 45. [aMCK, JW]
Kendler, K. S. & Greenspan, R. J. (in press) The nature of genetic influences on
behavior: Lessons from “simpler” organisms. American Journal of Psychiatry.
Kendler, K. S., Heath, A. C., Martin, N. G. & Eaves, L. J. (1987) Symptoms of
anxiety and symptoms of depression: Same genes, different environments?
Archives of General Psychiatry 44(5):451 – 57. [JW]
Kendler, K. S. & Karkowski-Shurman, L. (1997) Stressful life events and genetic
liability to major depression: Genetic control of exposure to the environment.
Psychological Medicine 27:539 – 47. [rMCK]
Kendler, K. S., McGuire, M. T., Gruenberg, A. M., O’Hare, A., Spellman, M. &
Walsh, D. (1993) The Roscommon family study. Archives of General Psychiatry 50:527 – 40. [aMCK]
Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C. & Eaves, L. J. (1992)
Major depression and generalized anxiety disorder: Same genes, (partly)
different environments? General Psychiatry 49:228– 47. [aMCK]
Kendler, K. S., Prescott, C. A., Myers, J. & Neale, M. C. (2003) The structure of
genetic and environmental risk factors for common psychiatric and substance
use disorders in men and women. Archives of General Psychiatry 60:929 –37.
Kendler, S., Gatz, M., Gardner, C. D. & Pedersen, N. L. (2006) A Swedish national
twin study of lifetime major depression. American Journal of Psychiatry
163:109 – 14. [JAS]
Kennedy, J. L., Farrer, L. A., Andreasen, N. C., Mayeux, R. & St. George-Hyslop,
P. (2003) The genetics of adult-onset neuropsychiatric disease: Complexities
and conundra? Science 302:822 – 26. [BC]
Kennedy, R. F. (2005) Deadly immunity. Rolling Stone, June 20, 2005:RSS 977/
978. Available at:
deadly_immunity. [MAG]
Kessler, R. C. (1997) The effects of stressful life events on depression. Annual
Review of Psychology 48:191 – 214. [aMCK]
Kessler, R. C., Berglund, P., Demler, O., Jin, R. & Walters, E. E. (2005) Lifetime
prevalence and age-of-onset distributions of DSM-IV disorders in the National
Comorbidity Survey Replication. Archives of General Psychiatry 62:593 – 602.
Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M.,
Eshleman, S., Wittchen, H. U. & Kendler, K. S. (1994) Lifetime and 12-month
prevalence of DSM-III-R psychiatric disorders in the United States:
Results from the National Comorbidity Survey. Archives of General Psychiatry
51:8 – 19. [JCW]
Killeen, P. R. & Nash, M. R. (2003) The four causes of hypnosis. International
Journal of Clinical and Experimental Hypnosis 51(3):195 – 231. [JW]
Kimura, M. (1983) The neutral theory of molecular evolution. Cambridge University
Press. [aMCK]
Kingham, M. & Gordon, H. (2004) Aspects of morbid jealousy. Advances in Psychiatric Treatment 10:207 – 15. [JAE]
Kirk, K. M., Blomberg, S. P., Duffy, D. L., Heath, A. C., Owens, I. F. P. & Martin,
N. M. (2001) Natural selection and quantitative genetics of life history traits in
Western women: A twin study. Evolution 55:423 – 35. [OM]
Kirley, A., Hawi, Z., Daly, G., McCarron, M., Mullins, C., Millar, N., Waldman, I.,
Fitzgerald, M. & Gill, M. (2002) Dopaminergic system genes in ADHD:
Toward a biological hypothesis. Neuropsychopharmacology 27:609 – 19.
Kitano, H. (2004) Biological robustness. Nature Reviews Genetics 5(11):826 – 37.
Kitano, T., Liu, Y. H., Ueda, S. & Saitou, N. (2004) Human-specific amino acid
changes found in 103 protein-coding genes. Molecular Biology and Evolution
21:936 – 44. [BC]
Klein, S. L. (2003) Parasite manipulation of the proximate mechanisms that
mediate social behavior in vertebrates. Physiology and Behavior 79:441 – 49.
Krebs, J. R. & Davies, N. B. (1993) An introduction to behavioural ecology,
3rd edition. Blackwell Scientific. [JSP]
Kreitman, M. (2000) Methods to detect selection in populations with applications to
the human. Annual Review of Genomics and Human Genetics 1:539– 59.
Kretschmer, E. (1970) Physique and character, 2nd edition. Couper Square.
Krueger, R. F. & Piasecki, T. M. (2002) Toward a dimensional and psychometrically-informed approach to conceptualizing psychopathology. Behavior
Research and Therapy 40:485– 99. [rMCK]
Krueger, R. F. & Tackett, J. L. (2003) Personality and psychopathology: Working
toward the bigger picture. Journal of Personality Disorders 17(2):109 – 28.
Kruglyak, L. & Nickerson, D. A. (2001) Variation is the spice of life. Nature Genetics
27:234 – 36. [aMCK]
Kyl-Heku, L. & Buss, D. M. (1996) Tactics as units of analysis in personality: An
illustration using tactics of hierarchy negotiation. Personality and Individual
Differences 21:497 – 517. [DMB]
Laing, R. D. (1960) The divided self. Tavistock. [R-PB]
Lande, R. (1982) A quantitative genetic theory of life history evolution. Ecology
63:607 – 15. [aMCK]
Lapierre, Y. D. (1994) Schizophrenia and manic-depression: Separate illnesses or a
continuum? Canadian Journal of Psychiatry 39(9, Suppl. 2):S59 – S64. [JP]
Laumann, E. O., Paik, A. & Rosen, R. C. (1999) Sexual dysfunction in the United
States: Prevalence and predictors. Journal of the American Medical Association
281:537– 44. [AT]
Lauronen, E., Veijola, J., Isohanni, I., Jones, P. B., Nieminen, P. & Isohanni, M.
(2004) Links between creativity and mental disorder. Psychiatry 67:81– 98.
Leboyer, M., Slama, F., Siever, L. & Bellivier, F. (2005) Suicidal disorders: A
nosological entity per se? American Journal of Medical Genetics, C: Seminars
in Medical Genetics 133:3 –7. [MV]
Ledgerwood, L. G., Ewald, P. W. & Cochran, G. M. (2003) Genes, germs, and
schizophrenia: An evolutionary perspective. Perspectives in Biology and
Medicine 46:317 – 48. [rMCK]
Leff, J., Sartorius, N., Jablensky, A., Korten, A. & Ernberg, G. (1992) The
International Pilot Study of Schizophrenia: Five-year follow-up findings.
Psychological Medicine 22:131 – 45. [AP]
Legrand, L. N., Iacono, W. G. & McGue, M. (2005) Predicting addiction. American
Scientist 93(2):140 – 48. [EIK]
Lehrke, R. (1997) Sex linkage of intelligence: The X-factor. Praeger. [MV]
Lerner, I. M. (1954) Genetic homeostasis. John Wiley. [aMCK]
Levinson, D. F. (2006) The genetics of depression: A review. Biological Psychiatry
60:84– 92. [JJM]
Li, D., Collier, D. A. & He, L. (2006) Meta-analysis shows strong positive
association of the neuregulin 1 (NRG1) gene with schizophrenia. Human
Molecular Genetics 15:1995 – 2002. [rMCK]
Li, D. & He, L. (2006) Further clarification of the contribution of the tryptophan
hydroxylase (TPH) gene to suicidal behavior using systematic allelic and genotypic meta-analyses. Human Genetics 119:233 – 40. [MV]
Li, L.-L., Keverne, E. B., Aparicio, S. A., Ishino, F., Barton, S. C. & Surani, M. A.
(1999) Regulation of maternal behavior and offspring growth by paternally
expressed Peg3. Science 284:330– 33. [rMCK]
Lickliter, R. & Honeycutt, H. (2003) Developmental dynamics: Toward a
biologically plausible evolutionary psychology. Psychological Bulletin
129:819– 35. [NBA]
Linn, E. L. (1977) Verbal auditory hallucinations: Mind, self, and society. Journal of
Nervous and Mental Disease 164:8– 17. [R-PB]
Loganovsky, K. N. & Loganovskaja, T. K. (2000) Schizophrenia spectrum disorders
in persons exposed to ionizing radiation as a result of the Chernobyl accident.
Schizophrenia Bulletin 26:751 – 53. [rMCK]
Lohmueller, K. E., Pearce, C. L., Pike, M., Lander, E. S. & Hirschhorn, J. N. (2003)
Meta-analysis of genetic association studies supports a contribution of common
variants to susceptibility to common disease. Nature Genetics 33:177 – 82.
Longley, A. J. (2001) Depression is an adaptation. Archives of General Psychiatry
58:1085– 86. [aMCK]
Lopez Leon, S., Croes, E. A., Sayed-Tabatabaei, F. A., Claes, S., Van Broeckhoven,
C. & van Duijn, C. M. (2005) The dopamine D4 receptor gene 48-base-pairrepeat polymorphism and mood disorders: A meta-analysis. Biological
Psychiatry 57:999 – 1003. [rMCK]
Loranger, A. W. (1981) Genetic independence of manic-depression and
schizophrenia. Acta Psychiatrica Scandinavica 63:444– 52. [JP]
Lorenz, K. (1963/2002) On aggression. Routledge. [R-PB]
Lu, J., Tang, T., Tang, H., Huang, J., Shi, S. & Wu, C.-I. (2006) The accumulation
of deleterious mutations in rice genomes: A hypothesis on the cost of
domestication. Trends in Genetics 22:126 – 31. [BC]
Lubinski, L., Benbow, C. P., Webb, R. M. & Bleske-Rechek, A. (2006) Tracking
exceptional human capital over two decades. Psychological Science 17:194– 99.
Luce, M. F., Bettman, J. R. & Payne, J. W. (1997) Choice processing in emotionally
difficult decisions. Journal of Experimental Psychology: Learning, Memory,
and Cognition 23(2):384 –405. [JW]
Ludwig, A. M. (1995) The price of greatness: Resolving the creativity and madness
controversy. Guilford Press. [DN]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Luedi, P. P., Hartemink, A. J. & Jirtle, R. L. (2005) Genome-wide prediction of
imprinted murine genes. Genome Research 15:875 – 84. [SWG]
Lund, O. C. H., Tamnes, C. K., Moestue, C., Buss, D. M. & Vollrath, M.
(in press) Tactics of hierarchy negotiation. Journal of Research in
Personality. [DMB]
MacIntyre, D. J., Blackwood, D., Porteous, D. J., Pickard, B. S. & Muir, W. J. (2003)
Chromosomal abnormalities and mental illness. Molecular Psychiatry
8:275 –87. [arMCK]
Maier, W., Lichtermann, D., Minges, J., Hallmayer, J., Heun, R., Benkert, O. &
Levinson, D. F. (1993) Continuity and discontinuity of affective disorders
and schizophrenia: Results of a controlled family study. Archives of General
Psychiatry 50(11):871 – 83. [JP]
Malaspina, D. (2001) Paternal factors and schizophrenia risk: De novo mutations
and imprinting. Schizophrenia Bulletin 27:379 – 93. [BC]
Malaspina, D., Corcoran, C., Fahim, C., Berman, A., Harkavy-Friedman, J.,
Yale, S., Goetz, D., Goetz, R., Harlap, S. & Gorman, J. (2002) Paternal age
and sporadic schizophrenia: Evidence for de novo mutations. American
Journal of Medical Genetics 114:299– 303. [rMCK]
Malaspina, D., Harlap, S., Fennig, S., Heiman, D., Noahon, D., Feldman, D. &
Susser, E. S. (2001) Advancing paternal age and the risk of schizophrenia.
Archives of General Psychiatry 58:361 – 67. [arMCK]
Mani, G. S., Clarke, B. C. & Shelton, P. R. (1990) A model of quantitative traits
under frequency-dependent balancing selection. Proceedings of the Royal
Society of London, Series B 240:15 – 28. [arMCK]
Mann, J. J. (2003) Neurobiology of suicidal behaviour. Nature Reviews
Neuroscience 4:819– 28. [MV]
Marshall, B. J., Goodwin, C. S., Warren, J. R., Murray, R., Blincow, E. D.,
Blackbourn, S. J., Phillips, M., Waters, T. E. & Sanderson, C. R. (1988)
Prospective double-blind trial of duodenal-ulcer relapse after eradication
of Campylobacter pylori. Lancet 2(8626 –8627):1437– 42. [rMCK]
Marušič, A. (2005) History and geography of suicide: Could genetic risk factors
account for the variation in suicide rates? American Journal of Medical
Genetics, C: Seminars in Medical Genetics 133:43 – 47. [MV]
Mather, K. (1974) Non-allelic interaction in continuous variation of randomly
breeding populations. Heredity 32:414 – 19. [aMCK]
Mattay, V. S., Goldberg, T. E., Fera, F., Hariri, A. R., Tessitore, A., Egan, M. F.,
Kolachana, B., Callicott, J. H. & Weinberger, D. R. (2003) Catechol
O-methyltransferase val158-met genotype and individual variation in the brain
response to amphetamine. Proceedings of the National Academy of Sciences
USA 100(10):6186– 91. [DCA]
Max, J. E., Koele, S. L., Smith, W. L., Sato, Y., Lindgren, S. D. & Robin, D. A.
(1998) Psychiatric disorders in children and adolescents after severe traumatic
brain injury: A controlled study. Journal of the American Academy of Child and
Adolescent Psychiatry 37:832 – 40. [aMCK]
Maynard Smith, J. (1982) Evolution and the theory of games. Cambridge University
Press. [JSP]
Mayo, O. (2004) Interaction and quantitative trait loci. Australian Journal of
Experimental Agriculture 44:1135 – 40. [OM]
Mayo, O., Bürger, R. & Leach, C. R. (1990) The heritability of fitness: Some single
gene models. Theoretical and Applied Genetics 79:278 – 84. [OM]
Maziade, M., Caron, C., Cote, R., Boutin, P. & Thivierge, J. (1990a) Extreme
temperament and diagnosis. Archives of General Psychiatry 47:477 –84. [EIK]
Maziade, M., Caron, C., Cote, R., Merette, C., Bernier, H., Laplante, B.,
Boutin, P. & Thievierge, J. (1990b) Psychiatric status of adolescents who had
extreme temperaments at age 7. American Journal of Psychiatry
147(11):1531 – 36. [EIK]
McAdams, D. P. & Pals, J. L. (2006) A new big five: Fundamental
principles for an integrative science of personality. American Psychologist
61:204 – 17. [JW]
McArdle, P. (2004) Attention-deficit hyperactivity disorder and life-span development. British Journal of Psychiatry 184:468 –69. [EIK]
McCrone, J. (2003) Darwinian medicine. Lancet Neurology 2:516. [aMCK, AT]
McDonald, C., Marshall, N., Sham, P. C., Bullmore, E. T., Schulze, K., Chapple, B.,
Bramon, E., Filbey, F., Quraishi, S., Walshe, M. & Murray, R. M. (2006)
Regional brain morphometry in patients with schizophrenia or bipolar disorder
and their unaffected relatives. American Journal of Psychiatry 163:478– 87.
McGrath, J. J. (2005) Myths and plain truths about schizophrenia epidemiology:
The NAPE lecture 2004. Acta Psychiatrica Scandinavica 111(1):4 –11. [JJM]
(2006) Variations in the incidence of schizophrenia: Data versus dogma.
Schizophrenia Bulletin 32(1):195 –97. [JJM]
McGrath, J. J., Hearle, J., Jenner, L., Plant, K., Drummond, A. & Barkla, J. M.
(1999) The fertility and fecundity of patients with psychosis. Acta Psychiatrica
Scandinavica 99:441 – 46. [aMCK]
McGrath, J. J., Saha, S., Welham, J., El Saadi, O., MacCauley, C. & Chant, D.
(2004) A systematic review of the incidence of schizophrenia: The distribution
of rates and the influence of sex, urbanicity, migrant status and methodology.
BMC Medicine 2(1):13. [JJM]
McGuffin, P., Rijsdijk, F., Andrew, M., Sham, P. C., Katz, R. & Cardno, A. (2003)
The heritability of bipolar affective disorder and the genetic relationship to
unipolar depression. Archives of General Psychiatry 60:497 – 502. [aMCK]
McGuire, M. T., Marks, I., Nesse, R. M. & Troisi, A. (1992) Evolutionary biology: A
basic science for psychiatry? Acta Psychiatrica Scandinavica 86(2):89 – 96.
McGuire, M. T. & Troisi, A. (1998) Darwinian psychiatry. Oxford University Press.
McKusick, V. A. (1998) Mendelian inheritance in man: Catalogs of human genes and
genetic disorders, 12th edition. Johns Hopkins Press. [aMCK]
Mealey, L. (1995) The sociobiology of sociopathy: An integrated evolutionary
model. Behavioral and Brain Sciences 18:523 – 99. [aMCK, DRW]
Merilä, J. & Sheldon, B. C. (1999) Genetic architecture of fitness and nonfitness
traits: Empirical patterns and development of ideas. Heredity 83:103 – 109.
Metzgar, D. & Wills, C. (2000) Evidence for the adaptive evolution of mutation
rates. Cell 101(6):581 – 84. [rMCK, JW]
Meyers, B. A. & Pueschel, S. M. (1991) Psychiatric disorders in persons with Down
syndrome. Journal of Nervous and Mental Disorders 179:609– 13. [rMCK]
Mill, J., Richards, S., Knight, J., Curran, S., Taylor, E. & Asherson, P. (2004)
Haplotype analysis of SNAP-25 suggests a role in the aetiology of ADHD.
Molecular Psychiatry 9:801– 10. [SWG]
Miller, G. F. (1997) Protean primates: The evolution of adaptive unpredictability in
competition and courtship. In: Machiavellian intelligence II: Extensions and
evaluations, ed. A. Whiten & R. W. Byrne. Cambridge University Press.
(2000a) The mating mind. Doubleday. [aMCK]
(2000b) Mental traits as fitness indicators: Expanding evolutionary psychology’s
adaptationism. In: Evolutionary perspectives on human reproductive behavior,
ed. D. LeCroy & P. Moller, pp. 62 – 74. Annals of the New York Academy of
Sciences, vol. 907. New York Academy of Sciences. [aMCK]
(2000c) Sexual selection for indicators of intelligence. In: Novartis Foundation
Symposium 233: The nature of intelligence, ed. G. Bock, J. Goode & K. Webb,
pp. 260 – 75. Wiley.
(2005) Vaginismus as an adaptive anti-copulation defence. Paper presented at the
annual meeting of the Human Behavior and Evolution Society, Austin, TX,
June 14, 2005. [rMCK]
Miller, G. F. & Todd, P. M. (1998) Mate choice turns cognitive. Trends in Cognitive
Sciences 2:190– 98. [aMCK]
Moffitt, T. E., Caspi, A., & Rutter, M. (2006) Measured gene-environment
interactions in psychopathology: Concepts, research strategies, and
implications for research, intervention, and public understanding of
genetics. Perspectives on Psychological Science 1:5– 27. [MV]
Mohr, C., Krummenacher, P., Landis, T., Sandor, P. S., Fathi, M. & Brugger, P.
(2005) Psychometric schizotypy modulates levodopa effects on lateralized
lexical decision performance. Journal of Psychiatric Research 39:241 – 50. [BC]
Monroe, S. M. & Simons, A. D. (1991) Diathesis-stress theories in the context of life
stress research: Implications for the depressive disorders. Psychological
Bulletin 110:406– 25. [aMCK]
Montague, P. R., Dayan, P. & Sejnowski, T. J. (1996) A framework for mesencephalic dopamine systems based on predictive Hebbian learning. Journal of
Neuroscience 16(5):1936 – 47. [JW]
Moore, B. R. (2004) The evolution of learning. Biological Reviews of the Cambridge
Philosophical Society 79:301 – 35. [JW]
Moriarty, P. J., Lieber, D., Bennett, A., White, L., Parrella, M., Harvey, P, D. &
Davis, K. L. (2001) Gender differences in poor outcome patients with lifelong
schizophrenia. Schizophrenia Bulletin 27:103– 13. [BC]
Morselli, E. (1881) Suicide: An essay on comparative moral statistics. Kegan Paul.
Mottron, L., Dawson, M., Soulieres, I., Hubert, B. & Burack, J. (2006)
Enhanced perceptual functioning in autism: An update, and eight principles of
autistic perception. Journal of Autism and Developmental Disorders 36:27– 43.
Mousseau, T. A., Sinervo, B. & Endler, J. (2000) Adaptive genetic variation in the
wild. Oxford University Press. [JP]
Mowat, R. R. (1966) Morbid jealousy and murder: A psychiatric study of morbidly
jealous murders at Broadmoor. Tavistock. [JAE]
Moynihan, R., Heath, I. & Henry, D. (2002) Selling sickness: The pharmaceutical
industry and disease mongering. British Medical Journal 324:886– 90. [AT]
Muglia, P., Petronis, A., Mundo, E., Lander, S., Cate, T. & Kennedy, J. L. (2002)
Dopamine D4 receptor and tyrosine hydroxylase genes in bipolar disorder:
Evidence for a role of DRD4. Molecular Psychiatry 7:860 – 66. [BC]
Munafo, M. R., Clark, T. G., Moore, L. R., Payne, E., Walton, R. & Flint, J. (2003)
Genetic polymorphisms and personality in healthy adults: A systematic review
and meta-analysis. Molecular Psychiatry 8:471– 84. [rMCK]
Nachman, M. W. (2002) Variation in recombination rate across the genome:
Evidence and implications. Current Opinion in Genetics and Development
12:657 – 63. [rMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Nachman, M. W. & Crowell, S. L. (2000) Estimate of the mutation rate per
nucleotide in humans. Genetics 156(1):297 – 304. [aMCK, JW]
Nanko, S. & Moridaira, N. S. (1993) Reproductive rates in schizophrenic
outpatients. Acta Psychiatrica Scandinavica 87:400 – 04. [aMCK]
Narrow, W. E., Rae, D. S., Robins, L. N. & Regier, D. A. (2002) Revised prevalence
estimates of mental disorders in the United States: Using a clinical significance
criterion to reconcile 2 surveys’ estimates. Archives of General Psychiatry
59:115 – 23. [JCW]
National Institute of Mental Health (1998) Genetics and Mental Disorders: Report
of the National Institute of Mental Health’s Genetic Workgroup. NIH
Publication No. 98 – 4268. [aMCK]
Nesse, R. M. (1991) Psychiatry. In: The sociobiological imagination, ed.
M. Maxwell. SUNY Press. [JW]
(2000) Is depression an adaptation? Archives of General Psychiatry 57:14– 20.
(2004) Cliff-edged fitness functions and the persistence of schizophrenia.
Behavioral and Brain Sciences 27(6):862 –63. [BC]
Nesse, R. M. & Berridge, K. C. (1997) Psychoactive drug use in evolutionary
perspective. Science 278:63 – 66. [aMCK]
Nesse, R. M. & Williams, G. C. (1994) Why we get sick: The new science of
Darwinian medicine. Times Books. [aMCK]
Nettle, D. (2001) Strong imagination: Madness, creativity and human nature.
Oxford University Press. [BC, DN]
(2004) Evolutionary origins of depression: A review and reformulation. Journal of
Affective Disorders 81:91– 102. [NBA, MB, DN]
(in press) Schizotypy and mental health amongst poets, visual artists and
mathematicians. Journal of Research in Personality [BC, DN]
Nettle, D. & Clegg, H. (2006) Schizotypy, creativity and mating success in humans.
Proceedings of the Royal Society of London, B: Biological Sciences 273:611 –15.
[BC, DN]
Nigg, J. T., John, O. P., Blaskey, L. G., Huang-Pollock, C. L., Willcutt, E. G.,
Hinshaw, S. P. & Pennington, B. (2002) Big five dimensions and ADHD
symptoms: Links between personality traits and clinical symptoms. Journal of
Personality and Social Psychology 83(2):451 –69. [EIK]
Nowakowska, C., Strong, C. M., Santosa, S., Wang, P. W. & Ketter, T. A. (2004)
Temperamental commonalities and differences in euthymic mood disorder
patients, creative controls, and healthy controls. Journal of Affective Disorders
85:207 – 15. [DN]
Online Mendelian Inheritance of Man [OMIM] (TM). McKusick-Nathans Institute
for Genetic Medicine, Johns Hopkis University (Baltimore, MD) and National
Center for Biotechnology Information, National Library of Medicine
(Bethesda, MD), April 10, 2006. World Wide Web URL: http:// [aMCK]
O’Reilly, T., Dunbar, R. & Bentall, R. (2001) Schizotypy and creativity: An
evolutionary connection? Personality and Individual Differences 31:1067 –78.
Otake, M. (1996) Threshold for radiation-related severe mental retardation in
prenatally exposed A-bomb survivors: A reanalysis. International Journal of
Radiation Biology 70:755 – 63. [rMCK]
Otto, S. P. (2000) Detecting the form of selection from DNA sequence data. Trends
in Genetics 16:526 – 29. [aMCK]
Oyama, S., Griffiths, P. E. & Gray, R. D. (2001) Cycles of contingency: Developmental systems and evolution. MIT Press. [NBA]
Palmatier, M. A., Kang, A. M. & Kidd, K. K. (1999) Global variation in the
frequencies of functionally different catechol-O-methyltransferase alleles.
Biological Psychiatry 46(4):557– 67. [JJM]
Parker, G. A. (1984) Evolutionarily stable strategies. In: Behavioural ecology: An
evolutionary approach, 2nd edition, ed. J. R. Krebs & N. B. Davies. Blackwell
Scientific. [JSP]
Pauls, D. L. (2003) An update on the genetics of Gilles de la Tourette syndrome.
Journal of Psychosomatic Research 55(1):7 – 12. Available at: http://¼pubmed&cmd¼Search&
itool¼pubmed_Abstract&term¼%22Pauls þ DL%22%5BAuthor%5D. [DRW]
Pearl, J. (2000) Causality: Models, reasoning, and inference. Cambridge University
Press. [JW]
Pedersen, C. B. & Mortensen, P. B. (2001) Evidence of a dose-response
relationship between urbanicity during upbringing and schizophrenia risk.
Archives of General Psychiatry 58:1039 –46. [R-PB]
Peters, S. U., Beaudet, A. L., Madduri, N. & Bacino, C. A. (2004) Autism in
Angelman syndrome: Implications for autism research. Clinical Genetics
66:530 – 36. [BC]
Pezawas, L., Meyer-Lindenberg, A., Drabant, E. M., Verchinski, B. A., Munoz, K. E.,
Kolachana, B. S., Egan, M. F., Mattay, V. S., Hariri, A. R. & Weinberger, D. R.
(2005) 5-HTTLPR polymorphism impacts human cingulate-amygdala
interactions: A genetic susceptibility mechanism for depression. Nature
Neuroscience 8(6):828 – 34. [DCA]
Pickover, C. A. (1999) Strange brains and genius: The secret lives of scientists and
madmen. Harper Perennial. [MAG]
Piven, J. & Palmer, P. (1999) Psychiatric disorder and the broad autism phenotype:
Evidence from a family study of multiple-incidence autism families. American
Journal of Psychiatry 156:557– 63. [aMCK]
Plomin, R. (1999) Genetic research on general cognitive ability as a model for mild
mental retardation. International Review of Psychiatry 11:34– 36. [aMCK]
Plomin, R., Corley, R., DeFries, J. C. & Fulker, D. W. (1990) Individual
differences in television viewing in early childhood: Nature as well as nurture.
Psychological Science 1:371– 77. [aMCK]
Plomin, R., DeFries, J. C., McClearn, G. E. & McGuffin, P. (2001) Behavioral
genetics, 4th edition. Worth. [aMCK]
Polimeni, J. & Reiss, J. P. (2002) How shamanism and group selection may reveal
the origins of schizophrenia. Medical Hypotheses 58:244 – 48. [aMCK]
(2004) Evolution and schizophrenia. In: Progress in schizophrenia research, ed.
J. E. Pletson, pp. 1 – 21. Nova Science. [JP]
Polimeni, J., Reiss, J. P. & Sareen, J. (2005) Could obsessive-compulsive disorder
have originated as a group-selected adaptive trait in traditional societies?
Medical Hypotheses 65:655 – 64. [JP]
Potash, J. B., Chiu, Y. F., MacKinnon, D. F., Miller, E. B., Simpson, S. G.,
McMahon, F. J., McInnis, M. G. & DePaulo, J. R., Jr. (2003) Familial
aggregation of psychotic symptoms in a replication set of 69 bipolar disorder
pedigrees. American Journal of Medical Genetics, B: Neuropsychiatric
Genetics 116(1):90– 97. [JP]
Preisig, M., Ferrero, F. & Malafosse, A. (2005) Monoamine oxidase a and
tryptophan hydroxylase gene polymorphisms: Are they associated in bipolar
disorder? American Journal of Pharmacogenetics 5:45 – 52. [rMCK]
Preti, A. & Miotto, P. (1997) Creativity, evolution and mental illnesses. Journal
of Memetics: Evolutionary Models of Information Transmission, 1.
Available at
(2005) Genetics, perinatal insult, and schizophrenia: The mechanism
underlying an increased prevalence of perinatal complications among
individuals with a diagnosis of schizophrenia? Current Psychiatry Reviews
1:139– 50. [AP]
Preti, A., Miotto, P. & Zen, T. (1998) Perinatal complications, genetic risk and
schizophrenia. Acta Psychiatrica Scandinavica 97:381– 83. [AP]
Price, J. S., Gardner, R. & Erickson, M. (2004) Can depression, anxiety and
somatisation be understood as appeasement displays? Journal of Affective
Disorders 79:1 – 11. Available at:
default.asp?k¼473. [JSP]
Price, J. S., Sloman, L., Gardner, R., Gilbert, P. & Rhode, P. (1994) The social
competition hypothesis of depression. British Journal of Psychiatry
164:309– 15. [rMCK]
Price, J. S. & Stevens, A. (1999) An evolutionary approach to psychiatric disorders:
Group splitting and schizophrenia. In: The evolution of the psyche, ed.
D. Rosen & M. Luebbert. Greenwood. Available at: htpp://www.johnprice.¼478. [JSP]
Price, T. & Schluter, D. (1991) On the low heritability of life-history traits.
Evolution 45:853 – 61. [aMCK]
Pritchard, J. K. (2001) Are rare variants responsible for susceptibility to complex
diseases? American Journal of Human Genetics 69:124 –37. [aMCK]
Prokosch, M., Yeo, R. & Miller, G. F. (2005) Intelligence tests with higher
g-loadings show higher correlations with body symmetry: Evidence for a
general fitness factor mediated by developmental stability. Intelligence
33:203 – 13. [aMCK]
Propping, P. (1983) Genetic disorders presenting as “schizophrenia”: Karl
Bonhoeffer’s early view of the psychoses in the light of medical genetics.
Human Genetics 65:1 –10. [aMCK]
Prout, T. (1999) How well does opposing selection maintain variation? In:
Evolutionary genetics from molecules to morphology, ed. R. S. Singh &
C. B. Krimbas, pp. 369 – 92. Cambridge University Press. [arMCK]
Pulver, A. E., McGrath, J. A., Liang, K. Y., Lasseter, V. K., Nestadt, G. &
Wolyniec, P. S. (2004) An indirect test of the new mutation hypothesis
associating advanced paternal age with the etiology of schizophrenia. American
Journal of Medical Genetics. Part B: Neuropsychiatric Genetics 124:6– 9.
Rahman, Q. (2005) The neurodevelopment of human sexual orientation.
Neuroscience and Biobehavioral Reviews 29(7):1057 – 66. Available at:¼pubmed&cmd¼
Author%5D. [DRW]
Ramus, F. (2004) Neurobiology of dyslexia: A reinterpretation of the data. Trends in
Neurosciences 27(12):720 – 26. [JW]
Randall, P. L. (1983) Schizophrenia, abnormal connection, and brain evolution.
Medical Hypotheses 10:247 – 80. [JP]
(1998) Schizophrenia as a consequence of brain evolution. Schizophrenia
Research 30:143 – 48. [JP]
Rao, V. & Lyketsos, C. (2000) Neuropsychiatric sequelae of traumatic brain injury.
Psychosomatics 41:95 – 103. [aMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Raybould, R., Green, E. K., MacGregor, S., Gordon-Smith, K., Heron, J., Hyde, S.,
Caesar, S., Nikolov, I., Williams, N., Jones, L., O’Donovan, M. C., Owen, M. J.,
Jones, I., Kirov, G. & Craddock, N. (2005) Bipolar disorder and
polymorphisms in the dysbindin gene (DTNBP1). Biological Psychiatry
57(7):696 – 701. [JJM]
Reed, E. W. (1971) Mental retardation and fertility. Social Biology 18(Suppl. 1):
42 –49. [aMCK]
Regier, D. A., Kaelbar, C. T., Rae, D. S., Farmer, M. E., Knauper, B., Kessler, R. C.
& Norquist, G. S. (1998) Limitations of diagnostic criteria and assessment
instruments for mental disorders: Implications for research and policy.
Archives of General Psychiatry 55:109 – 15. [JCW]
Reich, D. E. & Lander, E. S. (2001) On the allelic spectrum of human diseases.
Trends in Genetics 17:502 – 10. [aMCK]
Reichert, S. E. (1998) Game theory and animal contests. In: Game theory and animal
behavior, ed. L. A. Dugatkin & H. K. Reeve. Oxford University Press. [JSP]
Reik, W., Maher, E. R., Morrison, P. J., Harding, A. E. & Simpson, J. A. (1993) Age
at onset in Huntington’s disease and methylation at D4S95. Journal of Medical
Genetics 30:185 – 88. [BC, rMCK]
Reynolds, I. J., Malaiyandi, L. M., Coash, M. & Rintoul, G. L. (2004) Mitochondrial
trafficking in neurons: A key variable in neurodegeneration? Journal of
Bioenergetics and Biomembranes 36(4):283– 86. [JJM]
Rice, D. P., Kelman, S. & Miller, L. S. (1992) The economic burden of mental
illness. Hospital & Community Psychiatry 43:1227 – 32. [aMCK]
Rice, W. R. & Holland, B. (1997) The enemies within: Intergenomic conflict,
interlocus contest evolution (ICE), and the intraspecific Red Queen.
Behavioral Ecology and Sociobiology 41:1 –10. [SWG, rMCK, MV]
Richerson, P. J. & Boyd, R. (2004) Not by genes alone: How culture transformed
human evolution. University of Chicago Press. [JW]
Ridley, M. (1996) Evolution, 2nd edition. Blackwell. [aMCK]
(2000) Mendel’s demon: Gene justice and the complexity of life. Orion. [arMCK]
Rimmer, J. & Jacobsen, B. (1976) Differential fertility of adopted schizophrenics
and their half-siblings. Acta Psychiatrica Scandinavica 54:161 – 66. [arMCK]
Risch, N. (1990) Linkage strategies for genetically complex traits. I. Multilocus
models. American Journal of Human Genetics 46:222– 28. [aMCK]
Roff, D. A. (1997) Evolutionary quantitative genetics. Chapman & Hall. [arMCK]
Roff, D. A. & Mousseau, T. A. (1987) Quantitative genetics and fitness. Heredity
58:103 – 18. [aMCK]
Ropar, D. & Mitchell, P. (2002) Shape constancy in autism: The role of prior
knowledge and perspective cues. Journal of Child Psychology and Psychiatry
43(5):647 – 53. [MAG]
Rosenthal, N. (2006) Winter blues: Everything you need to beat seasonal affective
disorder, revised edition. Guilford Press. [JAS]
Ross, C. A. & Pam, A. (1995) Pseudoscience in biological psychiatry: Blaming the
body. Wiley. [aMCK]
Rudan, I., Rudan, D., Campbell, H., Carothers, A., Wright, A., Smolej-Narancic, N.,
Janicijevic, B., Jin, L., Chakraborty, R., Deka, R. & Rudan, P. (2003a)
Inbreeding and risk of late onset complex disease. Journal of Medical
Genetics 40:925 – 32. [aMCK]
Rudan, I., Smolej-Narancic, N., Campbell, H., Carothers, A., Wright, A.,
Janicijevic, B. & Rudan, P. (2003b) Inbreeding and the genetic complexity
of human hypertension. Genetics 163:1011 – 21. [aMCK]
Saha, S., Chant, D., Welham, J. & McGrath, J. J. (2005) A systematic review of the
prevalence of schizophrenia. PLoS Medicine 2:413 – 33. [Online at: PLoS
Medicine 2:e141. DOI:10.1371/journal.pmed.0020141.] [rMCK, AP]
Saito, T., Guan F., Papolos, D. F., Lau, S., Klein, M., Fann, C. S. & Lachman, H. M.
(2001) Mutation analysis of SYNJ1: A possible candidate gene for chromosome
21q22-linked bipolar disorder. Molecular Psychiatry 6:387– 95. [BC]
Samoilov, M., Arkin, A. & Ross, J. (2001) On the deduction of chemical reaction
pathways from measurements of time series of concentrations. Chaos
11(1):108 – 14. [JW]
Sandberg, R., Yasuda, R., Pakratz, D. G., Carter, T. A., Del Rio, J. A., Wodicka, L.,
Mayford, M., Lockhart, D. J. & Barlow, C. (2000) Regional and strain-specific
gene expression mapping in the adult mouse brain. Proceedings of the National
Academy of Sciences USA 97:11038 – 43. [aMCK]
Sanjuan, J., Tolosa, A., Gonzalez, J. C., Aguilar, E. J., Perez-Tur, J., Najera, C.,
Molto, M. D. & Frutos, R. (2006) Association between FOXP2 polymorphisms
and schizophrenia with auditory hallucinations. Psychiatric Genetics 16:67 – 72.
Sankaranarayanan, K. (2001) Estimation of the hereditary risks of exposure to
ionizing radiation: History, current status, and emerging perspectives. Health
Physics 80:363 – 69. [aMCK]
Saugstad, L. & Ödegard, Ö. (1986) Inbreeding and schizophrenia. Clinical Genetics
30:261 – 75. [aMCK]
Schoenhuber, R. & Gentilini, M. (1988) Anxiety and depression after mild head
injury: A case control study. Journal of Neurology, Neurosurgery, and
Psychiatry 51:722 – 24. [aMCK]
Schuldberg, D. (2000) Six subclinical spectrum traits in normal creativity.
Creativity Research Journal 13:5 – 16. [DN]
Seeman, M. V. (1979) Pathological jealousy. Psychiatry 42:351 –61. [JAE]
Shah, A. & Frith, U. (1983) An islet of ability in autistic children: A research note.
Journal of Child Psychology and Psychiatry 24:613 – 20. [MAG]
(1993) Why do autistic individuals show superior performance on the block
design task? Journal of Child Psychology and Psychiatry 34:1351 –64. [MAG]
Shaner, A., Miller, G. & Mintz, J. (2004) Schizophrenia as one extreme of a sexually
selected fitness indicator. Schizophrenia Research 70:101 – 109. [arMCK]
Shepherd, M. (1961) Morbid jealousy: Some clinical and social aspects of a
psychiatric symptom. Journal of Mental Science 107:687 – 753. [JAE]
Sherman, J. A. (2001) Evolutionary origin of bipolar disorder (EOBD). Psycoloquy
12. Available at:
Silberg, J., Pickles, A., Rutter, M., Hewitt, J., Simonoff, E., Maes, H.,
Carbonneau, R., Murrelle, L., Foley, D. & Eaves, L. (1999) The influence
of genetic factors and life stress on depression among adolescent girls.
Archives of General Psychiatry 56:225 – 32. [rMCK]
Simeonova, D. I., Chang, K. D., Strong, C. & Ketter, T. A. (2005) Creativity in
familial bipolar disorder. Journal of Psychiatric Research 39:623 – 31. [EIK]
Sipos, A., Rasmussen, R., Harrison, G., Tynelius, P., Lewis, G., Leon, D. A. &
Gunnell, D. (2004) Paternal age and schizophrenia: A population based cohort
study. British Medical Journal 329:1070 – 75. [BC, arMCK]
Slater, E. T., Hare, E. H. & Price, J. S. (1971) Marriage and fertility of psychiatric
patients compared with national data. Social Biology 18(Suppl. 1):60 –73.
Smalley, S. L., Loo, S. K., Yang, M. H. & Cantor, R. M. (2005) Toward
localizing genes underlying cerebral asymmetry and mental health. American
Journal of Medical Genetics, Part B: Neuropsychiatric Genetics 135:79 – 84.
Smith, C. (1975) Quantitative inheritance. In: Textbook of human genetics, ed.
G. R. Fraser & O. Mayo, pp. 382 – 441. Blackwell Scientific. [OM]
Smoller, J. W. & Finn, C. T. (2003) Family, twin and adoption studies of bipolar
disorder. American Journal of Medical Genetics, C: Seminars in Medical
Genetics 123(1):48– 58. [JP]
Sobell, J. L., Mikesell, M. J. & McMurray, C. T. (2002) Genetics and
etiopathophysiology of schizophrenia. Mayo Clinical Proceedings 77:1068– 82.
Somnath, C. P., Janardhan Reddy, Y. C. & Jain, S. (2002) Is there a familial overlap
between schizophrenia and bipolar disorder? Journal of Affective Disorders
72(3):243 – 47. [JP]
Sonuga-Barke, E. J. (2005) Causal models of attention-deficit/hyperactivity
disorder: From common simple deficits to multiple developmental pathways.
Biological Psychiatry 57(11):1231– 38. [JW]
Spinks, R., Sandhu, H. K., Andreasen, N. C. & Philibert, R. A. (2004) Association of
the HOPA12 bp allele with a large X-chromosome haplotype and positive
symptom schizophrenia. American Journal of Medical Genetics, Part B:
Neuropsychiatric Genetics 127:20 – 27. [BC]
Srinivasan, T. N. & Padmavati, R. (1997) Fertility and schizophrenia: Evidence for
increased fertility in the relatives of schizophrenic patients. Acta Psychiatrica
Scandinavica 96:260– 64. [rMCK]
Sternbach, R. A. (1963) Congenital insensitivity to pain. Psychological Bulletin
60:252 – 64. [rMCK]
Stevens, A. & Price, J. S. (2000a) Evolutionary psychiatry. Routledge. [aMCK]
(2000b) Prophets, cults and madness. Duckworth. [JSP]
Stevens, J. R. & Hauser, M. D. (2004) Why be nice? Psychological constraints on
the evolution of cooperation. Trends in Cognitive Sciences 8:60 – 65. [AP]
Stopkova, P., Vevera, J., Paclt, I., Zukov, I. & Lachman, H. M. (2004) Analysis of
SYNJ1, a candidate gene for 21q22 linked bipolar disorder: A replication study.
Psychiatry Research 127:157 – 61. [BC]
Sullivan, P. F. (2005) The genetics of schizophrenia. PLoS Medicine 2(7):614 – 18.
e212. Epub Jul 26. [rMCK, JJM]
Sullivan, P. F., Kendler, K. S. & Neale, M. C. (2003) Schizophrenia as a complex
trait: Evidence from a meta-analysis of twin studies. Archives of General
Psychiatry 60:1187– 92. [aMCK]
Sullivan, R. J. & Allen, J. S. (2004) Natural selection and schizophrenia. Behavioral
and Brain Sciences 27:865 – 66. [aMCK]
Sunyaev, S., Ramensky, R., Koch, I., Lathe, W., Kondrashov, A. S. & Bork, P. (2001)
Prediction of deleterious human alleles. Human Molecular Genetics
10:591 – 97. [aMCK]
Swanson, W. J., Zhang, Z. H., Wolfner, M. F. & Aquadro, C. F. (2001) Positive
Darwinian selection drives the evolution of several female reproductive
proteins in mammals. Proceedings of the National Academy of Sciences USA
98:2509 –14. [SWG]
Sykes, B. (2001) The seven daughters of Eve: The science that reveals our genetic
ancestry. W. W. Norton. [AP]
Symons, D. (1989) A critique of Darwinian anthropology. Ethology and Sociobiology 10:131 –44. [aMCK]
Taylor, L., Faraone, S. V. & Tsuang, M. T. (2002) Family, twin, and adoption studies
of bipolar disease. Current Psychiatry Reports 4(2):130 – 33. [JP]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Terwilliger, J. D. & Weiss, K. M. (1998) Linkage disequilibrium mapping of
complex disease: Fantasy or reality? Current Opinion in Biotechnology
9:578– 94. [aMCK]
Thoma, R. J., Yeo, R. A., Gangestad, S. W., Lewine. J. & Davis, J. (2002) Fluctuating
asymmetry and the human brain. Laterality 7:45 – 58. [SWG]
Thomas, V. S., Darvesh, S., MacKnight, C. & Rockwood, K. (2001) Estimating the
prevalence of dementia in elderly people: A comparison of the Canadian
study of health and aging and national population health survey approaches.
International Psychogeriatrics 13:169 – 75. [aMCK]
Tiemann-Boege, I., Navidi, W., Grewal, R., Cohn, D., Eskenazi, B., Wyrobek, A. J.
& Arnheim, N. (2002) The observed human sperm mutation frequency cannot
explain the achondroplasia paternal age effect. Proceedings of the National
Academy of Sciences USA 99:14952 –57. [BC, rMCK]
Toichi, M., Kamio, Y., Okada, T., Sakihama, M., Youngstrom, E. A., Findling, R. L.
& Yamamoto, K. (2002) A lack of self-consciousness in autism. American
Journal of Psychiatry 159:1422 – 24. [MAG]
Tooby, J. (1982) Pathogens, polymorphism, and the evolution of sex. Journal of
Theoretical Biology 94:557 – 76. [rMCK]
Tooby, J. & Cosmides, L. (1990) On the universality of human nature and the
uniqueness of the individual: The role of genetics and adaptation. Journal of
Personality 58:17 – 67. [aMCK]
Torrey, E. F., Bowler, A. E., Taylor, E. H. & Gottesman, I. I. (1994) Schizophrenia
and manic-depressive disorder. Basic Books. (pp. 173 – 175) [JP]
Tóth, G., Buda, B. L. & Eiben, O. G. (2002) Contribution to the physique of women
with manic-depressive disorder in Hungary. Collegium Antropologicum
27:581 – 86. [JAS]
Treloar, S. A., Wicks, J., Nyholt, D. R., Montgomery, G. W., Bahlo, M., Smith, V.,
Dawson, G., Mackay, I. J., Weeks, D. E., Bennett, S. T., Carey, A., EwenWhite, K. R., Duffy, D. L., O’Connor, D. T., Barlow, D. H., Martin, N. G. &
Kennedy, S. H. (2005) Genomewide linkage study in 1,176 affected sister pair
families identifies a significant susceptibility locus for endometriosis on
chromosome 10q26. American Journal of Human Genetics 77:365 – 76. [OM]
Trivers, R. (1971) Evolution of reciprocal altruism. Quarterly Review of Biology
46:35– 57. [AP]
Trivers, R. L. (1974) Parent-offspring conflict. American Zoologist 14:249–64. [SWG]
Troisi, A. (2005) The concept of alternative strategies and its relevance to psychiatry
and clinical psychology. Neuroscience and Biobehavioral Reviews 29:159– 68.
Troisi, A. & McGuire, M. T. (2002) Darwinian psychiatry and the concept of mental
disorder. Neuroendocrinology Letters 23(Suppl. 4):31 – 38. [arMCK, AT]
Tunbridge, E. M., Harrison, P. J. & Weinberger, D. R. (2006) Catechol-omethyltransferase, cognition, and psychosis: Val(158) and beyond. Met and
Beyond. Biological Psychiatry 60:141 – 51. [JJM]
Turelli, M. & Barton, N. H. (2004) Polygenic variation maintained by balancing
selection: Pleiotropy, sex-dependent allelic effects, and G E interactions.
Genetics 166:1053 – 79. [aMCK]
Tuulio-Henriksson, A., Arajärvi, R., Partonen, T., Haukka, J., Varilo, T.,
Schreck, M., Cannon, T. & Lönnqvist, J. (2003) Familial loading associates
with impairment in visual span among healthy siblings of schizophrenia
patients. Biological Psychiatry 54:623 – 28. [TC]
Vaillant, G. (2003) Mental health. American Journal of Psychiatry 160:1272 – 84.
van Os, J., Krabbendam, L., Myin-Germeys, I. & Delespaul, P. (2005). The
schizophrenia envirome. Current Opinion in Psychiatry 18:141 – 45. [JJM]
Van Valen, L. (1973) A new evolutionary law. Evolutionary Theory 1:1– 30. [SWG]
Vauhkonen, K. (1968) On the pathogenesis of morbid jealousy. Kunnallispaino.
Venter, J. C., Adams, M. D., Myers, E. W., Li, P. W., Mural, R. J., Sutton, G. G.,
et al. [274 coauthors] (2001) The sequence of the human genome. Science
291:1304 – 51. [rMCK]
Verdoux, H., Geddes, J. R., Takei, N., Lawrie, S. M., Bovet, P., Eagles, J. M.,
Heun, R., McCreadie, R. G., McNeil, T. F., O’Callaghan, E., Stober, G.,
Willinger, M. U., Wright, P. & Murray, R. M. (1997) Obstetric complications
and age at onset in schizophrenia: An international collaborative meta-analysis
of individual patient data. American Journal of Psychiatry 154:1220 – 27.
Vogel, F. & Motulsky, A. G. (1997) Human genetics, 3rd edition. Springer-Verlag.
Vogel, H. P. (1979) Fertility and sibship size in a psychiatric patient population.
Acta Psychiatrica Scandinavica 60:483 – 503. [arMCK]
Vogels, A., De Hert, M., Descheemaeker, M. J., Govers, V., Devriendt, K.,
Legius, E., Prinzie, P. & Fryns, J. P. (2004) Psychotic disorders in Prader-Willi
syndrome. American Journal of Medical Genetics, Part A 127:238 – 43. [BC]
Voight, B. F., Kudaravalli, S., Wen, X. & Pritchard, J. K. (2006) A map of
recent positive selection in the human genome. PLoS Biology 4(3):e72.
Voracek, M. (2004) National intelligence and suicide rate: An ecological study of 85
countries. Personality and Individual Differences 37:543– 53. [MV]
(2005a) National intelligence, suicide rate in the elderly, and a threshold
intelligence for suicidality: An ecological study of 48 Eurasian countries.
Journal of Biosocial Science 37:721 – 40. [MV]
(2005b) The social ecology of intelligence and suicide in Belarus. Journal of Social
Psychology 145:613– 17. [MV]
(2006a) RE: Childhood cognitive performance and risk of mortality: A
prospective cohort study of gifted individuals [Letter]. American Journal of
Epidemiology 163:1161 – 62. [MV]
(2006b) Regional intelligence and suicide rate in Denmark. Psychological
Reports 98:671– 72. [MV]
(2006c) Smart and suicidal? The social ecology of intelligence and suicide in
Austria. Death Studies 30:471 – 85. [MV]
(2006d) Social ecology of intelligence and suicide in the United States. Perceptual
and Motor Skills 102:767 – 75. [MV]
Voracek, M., Fisher, M. L. & Marušič, A. (2003) The Finno-Ugrian suicide
hypothesis: Variation in European suicide rates by latitude and longitude.
Perceptual and Motor Skills 97:401 – 406. [MV]
Voracek, M. & Formann, A. K. (2004) Variation in European suicide rates is better
accounted for by latitude and longitude than by national percentage of FinnoUgrians and Type O blood: A rebuttal of Lester and Kondrichin (2004).
Perceptual and Motor Skills 99:1243 – 50. [MV]
Waddell, C. (1998) Creativity and mental illness: Is there a link? Canadian Journal
of Psychiatry 43:166 – 72. [BC, aMCK]
Wakefield, J. C. (1992) The concept of mental disorder: On the boundary between
biological facts and social values. American Psychologist 47:373 – 88. [arMCK,
(1994) Is the concept of mental disorder culturally relative? In: Controversial
issues in mental health, ed. S. Kirk & S. Einbinder, pp. 11– 17. Allyn and
Bacon. [JCW]
(1999a) Evolutionary versus prototype analysis of the concept of disorder. Journal
of Abnormal Psychology 108:374– 99. [NBA, JAE, aMCK, JCW]
(1999b) The measurement of mental disorder. In: A handbook for the study of
mental health, ed. A. V. Horwitz & T. L. Scheid, pp. 29 – 57. Cambridge
University Press. [JCW]
(2005) Biological function and dysfunction. In: The handbook of evolutionary
psychology, ed. D. M. Buss, pp. 878 –902. Wiley. [JAE, JCW]
Wakefield, J. C., Horwitz, A. V. & Schmitz, M. (2005a) Are we overpathologizing
social anxiety? Social phobia from a harmful dysfunction perspective.
Canadian Journal of Psychiatry 50:317 –19. [JCW]
(2005b) Social disadvantage is not mental disorder: Response to Campbell-Sills and
Stein. Canadian Journal of Psychiatry 50:324 – 26. [JCW]
Walker, E. F., Savoie, T. & Davis, D. (1994) Neuromotor precursors of
schizophrenia. Schizophrenia Bulletin 20:441– 51. [R-PB]
Waller, J. H. (1971) Differential reproduction: Its relation to I.Q. test score,
education and occupation. Social Biology 18:122 – 37. [aMCK]
Walshe, M., McDonald, C., Taylor, M., Zhao, J., Sham, P., Grech, A., Schulze, K.,
Bramon, E. & Murray, R. M. (2005) Obstetric complications in patients with
schizophrenia and their unaffected siblings. European Psychiatry 20:28–34. [AP]
Watson, P. & Andrews, P. (2002) Toward a revised evolutionary adaptationist
analysis of depression: The social navigation hypothesis. Journal of Affective
Disorders 72(1):1 –14. [arMCK, JW]
Wehr, T. A., Duncan, W. C., Sher, L., Aeschbach, D., Schwartz, P. J., Turner, E. H.,
Postolache, T. T. & Rosenthal, N. E. (2001) A circadian signal of change of
season in patients with seasonal affective disorder. Archives of General
Psychiatry 58:1002– 14. [JAS]
Weisfeld, G. E. (2004) Some ethological perspectives on the fitness consequences
and social emotional symptoms of schizophrenia. Behavioral and Brain
Sciences 27(6):867. [aMCK]
Weiss, G., Hechtman, L. Milroy, T. & Perlman, T. (1985) Psychiatric status of
hyperactives as adults: A controlled prospective 15-year follow-up of 63
hyperactive children. Journal of the American Academy of Child and
Adolescent Psychiatry 24:211 – 20. [EIK]
Weiss, K. M. & Clark, A. G. (2002) Linkage disequilibrium and the mapping of
complex human traits. Trends in Genetics 18:19 – 24. [aMCK]
Weissman, M. M., Bland, R. C., Canino, G. J., Faravelli, C., Greenwald, S., Hwu,
H. G., Joyce, P. R., Karamm, E. G., Lee, C. K., Lellouch, J., Lepine, J. P.,
Newman, S. C., Rubio-Stipec, M., Wells, J. E., Wickramaratne, P. J.,
Wittchen, H. & Yeh, E. K. (1996) Cross-national epidemiology of major
depression and bipolar disorder. Journal of the American Medical Association
276:293– 99. [aMCK]
Wen, X., Fuhrman, S., Michaels, G. S., Carr, D. B., Smith, S., Barker, J. L. &
Somogyi, R. (1998) Large-scale temporal gene expression mapping of central
nervous system development. Proceedings of the National Academy of Sciences
USA 95(1):334– 39. [JW]
Widiger, T. A. & Samuel, D. B. (2005) Diagnostic categories or dimensions? A
question for the Diagnostic and Statistical Manual of Mental Disorders,
5th edition. Journal of Abnormal Psychology 114:494– 504. [rMCK]
Willerman, L. & Cohen, D. B. (1990) Psychopathology. McGraw-Hill. [rMCK]
References/Keller & Miller: Resolving the paradox of heritable mental disorders
Williams, G. C. (1966) Adaptation and natural selection: A critique of some current
evolutionary thought. Princeton University Press. [aMCK]
Williams, J. (2005) Regulated turbulence in information. In: Convergence, ed.
W. Ostreng. Oslo Centre for Advanced Studies. Available at http://www.cas. [JW]
Williams, J. & Dayan, P. (2005) Dopamine, learning, and impulsivity: A biological
account of attention-deficit/hyperactivity disorder. Journal of Child and
Adolescent Psychopharmacology 15(2):160 – 79. [JW]
Williams, J. & Taylor, E. (2004) Dopamine appetite and cognitive impairment in
attention-deficit/hyperactivity disorder. Neural Plasticity 11(1):115 –32. [JW]
(2006) The evolution of hyperactivity, impulsivity and behavioural diversity.
Journal of the Royal Society Interface 3:399– 413. Available at http://¼article&id¼doi:10.1098/
rsif.2005.0102. [JW]
Wilson, D. R. (1998) Evolutionary epidemiology and manic depression. British
Journal of Medical Psychiatry 71(Pt. 4):367– 95. Available at: http://¼pubmed&cmd¼Retrieve&
docsum. [arMCK, AP, DRW]
(2001) Depression is an adaptation. Archives of General Psychiatry 58:1086 – 87.
Wilson, D. S. (1994) Adaptive genetic variation and human evolutionary
psychology. Ethology and Sociobiology 15:219 – 35. [aMCK]
Woody, E. & Claridge, G. (1977) Psychoticism and thinking. British Journal of
Social and Clinical Psychology 16:241– 48. [DN]
Wright, A. F., Charlesworth, B., Rudan, I., Carothers, A. & Campbell, H. (2003) A
polygenic basis for late-onset disease. Trends in Genetics 19:97 – 106. [arMCK]
Wright, A. F. & Hastie, N. D. (2001) Complex genetic diseases: Controversy over
the Croesus code. Genome Biology 2:1 – 8. [aMCK]
Wright, P., Nimgaonkar, V. L., Donaldson, P. T. & Murray, R. M. (2001)
Schizophrenia and HLA: A review. Schizophrenia Research 15:1– 12. [rMCK]
Wyckoff, G. J., Wang, W. & Wu, C. L. (2000) Rapid evolution of male reproductive
genes in the descent of man. Nature 403:304 – 309. [SWG]
Xu, J. Z., Pato, M. T., Dalla Torre, C., Medeiros, H., Carvalho, C., Basile, V. S.,
Bauer, A., Dourado, A., Valente, J., Soares, M. J., Macedo, A. A., Coelho, I.,
Ferreira, C. P., Azevedo, M. H., Macciardi, F., Kennedy, J. L. & Pato, C. N.
(2001) Evidence for linkage disequilibrium between the alpha 7-nicotinic
receptor gene (CHRNA7) locus and schizophrenia in Azorean families.
American Journal of Human Genetics 105:669 – 74. [SWG]
Xu, M., St Clair, D. & He, L. (2006) Meta-analysis of association between ApoE
epsilon4 allele and schizophrenia. Schizophrenia Research 84:228–35. [rMCK]
Yeo, R. A. & Gangestad, S. W. (1993) Developmental origins of human hand
preference. Genetica 89:281 – 96. [SWG, rMCK]
(1998) Developmental instability and phenotypic variation in neural
organization. In: The other side of the error term: Aging and development
as models in cognitive neuroscience, ed. N. Raz, pp. 1– 51. Elsevier.
Yeo, R. A., Gangestad, S. W., Edgar, C. & Thoma, R. J. (1999) The
evolutionary-genetic underpinnings of schizophrenia: The developmental
instability model. Schizophrenia Research 39(3):197– 206. [SWG, rMCK,
Yeo, R. A., Gangestad, S. W., Thoma, R. J., Shaw, P. K. & Repa, K. (1997)
Developmental instability and cerebral lateralization. Neuropsychology
11:552 – 61. [SWG]
Yin, L., Wang, J., Klein, P. S. & Lazar, M. A. (2006) Nuclear receptor Rev-erba is
a critical lithium-sensitive component of the circadian clock. Science
311:1002 – 1005. [JAS]
Ylisaukko-Oja, T., Alarcón, M., Cantor, R. M., Auranen, M., Vanhala, R.,
Kempas, E., von Wendt, L., Järvelä, I., Geschwind, D. H. & Peltonen,
L. (2006) Search for autism loci by combined analysis of Autism Genetic
Resource Exchange and Finnish families. Annals of Neurology 59:145 – 55.
Yoshikawa, T., Kikuchi, M., Saito, K., Watanabe, A., Yamada, K., Shibuya, H.,
Nankai, M., Kurumaji, A., Hattori, E., Ishiguro, H., Shimizu, H., Okubo, Y.,
Toru, M. & Detera-Wadleigh, S. D. (2001) Evidence for association of the
myo-inositol monophosphatase 2 (IMPA2) gene with schizophrenia in
Japanese samples. Molecular Psychiatry 6:202 – 10. [BC]
Zanarini, M. C., Williams, A. A., Lewis, R. E., Reich, R. B., Vera, S. C., Marino,
M. F., Levin, A., Yong, L. & Frankenburg, F. R. (1997) Reported
pathological childhood experiences associated with the development of
borderline personality disorder. American Journal of Psychiatry 154(8):
1101 –106. [JP]
Zechner, U., Wilda, M., Kehrer-Sawatzki, H., Vogel, W., Fundele, R. & Hameister,
H. (2001) A high density of X-linked genes for general cognitive ability: A
run-away process shaping human evolution? Trends in Genetics 17:697 – 701.
Zhang, J., Webb, D. M. & Podlaha O. (2002) Accelerated protein evolution
and origins of human-specific features: FOXP2 as an example. Genetics
162:1825 – 35. [BC]
Zhang, S. L. (1992) A study on effects of parent age, birth order, and mental
retardation of unknown etiology [in Chinese; English abstract]. Chung Hua
Shen Ching Shen Ko Tsa Chih 25:303 – 305. [aMCK]
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF