CU R RE N T D I R E CT I O NS IN P SYC H OL OGI C AL SC I EN C E
Developmental Instability and
Individual Variation in Brain
Development
Implications for the Origin of Neurodevelopmental
Disorders
Ronald A. Yeo,1 Steven W. Gangestad,1 and Robert J. Thoma2,3
1
Department of Psychology, University of New Mexico, 2MIND Institute, Albuquerque, New Mexico, and
3
Department of Psychiatry, School of Medicine, University of New Mexico
ABSTRACT—Research on the origin of neurodevelopmental
disorders has traditionally been pursued within a con-
strained, disorder-specific perspective. The developmental
instability (DI) model described here offers a broader ap-
proach based on the evolutionary genetics of normal
variation, reflecting our understanding that the processes
generating genetic diversity are not unique to any specific
disorder. The DI model helps account for shared features,
including atypical functional and anatomic asymmetries,
reduced general intellectual functioning, and complex
patterns of heritability, across different types of neural
variation. The model suggests research strategies that may
help illuminate the specific and unique causal factors
characterizing different types of neural variation.
KEYWORDS—developmental instability; neurodevelopmen-
tal disorders; genetics; evolution
In this article we outline a model for the origins of individual
differences in brain development based on the construct of de-
velopmental instability (DI). This approach was pioneered by
evolutionary theorists such as Waddington and Lerner, and de-
veloped in recent years by Gottesman, Markow, Woolf, and
others. DI is conceptualized as a vulnerability factor for varia-
tion in brain development across both the normal spectrum and
that of neurodevelopmental disorders such as schizophrenia and
dyslexia. The DI model may help address the following ques-
Address correspondence to Ronald A. Yeo, Department of Psychology,
University of New Mexico, Albuquerque, New Mexico 87122; e-mail:
ryeo@unm.edu.
Volume 16—Number 5 Copyright r 2007 Association for Psychological Science
tions: Why do neurodevelopmental disorders so often occur to-
gether (comorbidity)? Why are some observable (phenotypic)
features commonly shared across neurodevelopmental disor-
ders? Why are disorders with adverse outcomes and a strong
genetic component nevertheless maintained at relatively stable
levels in the population?
DEVELOPMENTAL INSTABILITY: PHENOTYPES AND
GENOTYPES
Diverse genetic and environmental stressors introduce ‘‘noise’’
or imprecision into developmental pathways, and natural
selection favors organisms capable of buffering the impact of
these factors on the developmental processes that combine
to produce an organism’s observable characteristics—that is,
its phenotype. Thus, organisms continually face the adaptive
problem of how to buffer themselves against such perturbations.
Because the specific factors causing DI are difficult to mea-
sure directly, researchers have identified two different types of
higher-level proxy variables, or markers, of DI. These markers
are in turn associated with measurable phenotypic variations, or
correlates, of DI. One marker is fluctuating asymmetry (FA):
deviations from symmetry on bilateral traits such as finger length
or ear width that are symmetric at a population level. Because
the same genes control development of the left and right sides
of the body, deviation from perfect symmetry reflects develop-
mental error creeping into growth processes. For example, left
and right ear length is equal at a population level, yet each of
us shows minor deviations from perfect symmetry. Researchers
often aggregate several fluctuating asymmetries into composite
scores that show moderate heritability (Mller, 2006). The
245
 Developmental Instability and Neurodevelopmental Disorders
second marker is a composite measure of minor physical
anomalies (MPAs), physical features caused by slow or disrupted
fetal growth rates. For example, hypertelorism (wide-spaced
eyes) reflects slow growth during a particular point in prenatal
development.
Measures of DI correlate with important traits in normal
populations. FA accounts for a substantial amount of variance in
general intellectual functioning; greater FA predicts lower in-
telligence, especially on intellectual tests tapping general
ability or ‘‘g’’ (Prokosch, Yeo, & Miller, 2005). The relationship
between DI and intelligence is not mediated by variation in brain
volumes but may be mediated by neural-processing speed, be-
cause faster processing during cognitive tasks has been linked
with reduced FA. Several studies (Yeo, Thoma, & Gangestad,
2002) have shown that individuals with greater DI show atypical
functional brain asymmetry, evidenced by atypical hand skill
(both left handedness and extreme right handedness) and
atypical cognitive asymmetry. Figure 1 shows the relationship
between variation in hand skill and DI (Yeo et al., 2002). Note
that moderate right handedness, the population mean, is asso-
ciated with the lowest level of DI. DI also predicts atypical an-
atomic brain asymmetry, including atypical asymmetry of
temporal lobe structures critical for language. DI appears to
be an important factor in human mate selection. Men with lower
FA (less DI) have more sexual partners and are preferred by
females during the fertile part of their menstrual cycle (Gan-
gestad, Thornhill, & Garver-Apgar, 2005). Other studies link
reduced FA with fewer infections, stronger immunity, and better
health (Mller, 2006).
1.00
Developmental Instability
0.50
0.00
−0.50
−1.50 −1.00 −0.50 0.00 0.50
Relative Hand Skill
Fig. 1. Hinged linear relationship between developmental instability and
hand skill on Annett’s Peg Moving task. Hand skill is represented as right-
minus-left latency (peg-moving speed), with extreme right-handedness on
the leftmost side, left-handedness toward the right side, and median rel-
ative hand skill at the low point of the figure. The DI score is a composite of
skeletal fluctuating asymmetry scores and minor physical anomalies. Each
data point represents the performance of groups of nine or more individ-
uals. See Yeo, Thoma, and Gangestad (2002) for additional details.
246
Foremost among the genetic influences on DI is the number
of genetic mutations an individual carries, referred to as one’s
mutation load. Estimates of the number of new, harmful muta-
tions per generation range from 1.6 to 3 (Prokosch et al., 2005).
Individuals may have up to several hundred mildly deleterious
mutations that will eventually be removed from the population
through natural selection. The selective advantage of low-FA
males in mating contexts may reflect a female preference
for their ‘‘advantageous genes.’’ DI may also be greater when
individuals have relatively more genetic loci where the alleles
are the same (homozygous)—as would be the case with
inbreeding—than when individuals have differing alleles
at many genetic loci. Environmental factors can also increase
DI; prenatal stress, toxin exposure, and illness are probably the
most important.
Various factors enable alleles that contribute to adverse out-
comes to persist in the human gene pool. First, the causes of new
mutations, such as ionizing radiation, are ongoing, where
removal of mutations from the gene pool occurs over generations.
Second, heterozygosity at a given locus cannot ‘‘breed true’’—
that is, some offspring of two heterozygous parents will inevitably
be homozygous (as for example, when one Tt genotype mates
with another Tt, one quarter of their offspring will be homozygous
TT and one quarter will be homozygous tt). Third, genetic
conflicts between members of a species may fuel persistent
antagonistic coevolution of genes. For example, a fetus’s genes
maximally benefit from a greater flow of nutrients from the
mother than would be optimal to maximize the mother’s fitness.
Just as football offenses and defenses coevolve over time through
persistent introduction of new adaptive variants, so too maternal
and fetal ‘‘strategies’’ coevolve through persistent introduction of
new genetic variants. Because new variants are not introduced
into all individuals simultaneously, this process helps maintain
genetic variation.
There need not be some hidden benefit (e.g., creativity) to
genes linked with neurodevelopmental disorders for the genes to
be passed on and maintained in the gene pool. Disadvantaged
alleles will always be present. The best that individuals can do
is to minimize the frequency of these alleles in their offspring
via careful mate selection. Hence, humans have evolved psy-
chological mechanisms for seeking mates with genomes less apt
to lead to atypical development.
As DI disrupts mechanisms that help maintain the integrity of
developmental processes, one consequence of greater DI is a
1.00 greater impact of environmental stressors. For example, Rose
(2005) compared genetically identical (monozygous) twin pairs
with high FA with monozygous twin pairs with low FA. Twin pairs
with higher FA were significantly less alike on a standard
personality test, and these results were replicated in an
independent sample. Similarly, high-FA preschoolers who were
exposed to family stress had more negative psychological
adjustment than low-FA children with equivalent stress expo-
sure (Rose, 2005). These reports suggest that greater DI leads to
Volume 16—Number 5
 Ronald A. Yeo, Steven W. Gangestad, and Robert J. Thoma
reduced buffering from environmental stress—and thus that
individuals with high DI are an ‘‘at risk’’ population. Although
data are lacking, the genetic factors underlying greater DI may
also permit a greater functional impact of single genes known to
be relevant for individual variation in behavior.
A MODEL FOR NEURODEVELOPMENTAL DISORDERS
We have proposed a two-factor DI model for neurodevelopmental
disorders (Yeo, Gangestad, Edgar, & Thoma, 1999). The first
factor refers to the generalized effects of DI. Mutations at
different loci throughout the genome (and other genetic factors)
contribute to DI (see Fig. 2). Although the specific mutations will
vary across individuals, they may well have a similar impact on
developmental processes, making the identification of specific
genes important for neurodevelopmental variation quite
difficult. Further, as the impact of DI is not specific to a given
developmental pathway, individuals with perturbed develop-
ment of one phenotypic feature may also demonstrate perturbed
development of other features. Hence, high DI leads to ‘‘corre-
lated atypicalities,’’ or comorbidity.
The second factor reflects genetic or environmental causes
that are specific to particular atypical outcomes. For example,
certain mutations or other genetic variants may affect only
Fig. 2. The idealized two-factor developmental instability (DI) model.
DI, or imprecise expression of developmental design, is caused by a number
of factors including mutations; maternal–fetal conflicts; and environmental
perturbations such as those due to pathogens, free radicals, biotoxins, and
nutritional stresses. DI may result in neuropathology, expressed as neuro-
developmental disorder, but the precise outcomes of DI vary. Specific de-
velopmental disorders may also be affected by factors specific to them (e.g.,
genes affecting neurotransmitters or receptors, precise timing of develop-
mental perturbation). DI likely interacts with specific factors to influence
outcomes (e.g., if DI is high, lower ‘‘doses’’ of specific factors may result in
dysfunction). Though we represent just two disorders here (schizophrenia
and attention-deficit/hyperactivity disorder, ADHD), the model applies to
all neurodevelopmental disorders. This version is idealized because causes
probably vary along a continuum rather than falling into two discrete classes
of factors.
Volume 16—Number 5
particular developmental processes and, in the context of high
DI, result in specific atypical outcomes. Gene-by-gene inter-
actions are expected to occur, such that, for individuals of high
DI, lower ‘‘doses’’ of specific liabilities are required to result in
the development of a disorder. Environmental factors may also
cause specific outcomes. Frequent examples may be environ-
mental stressors (e.g., infections) that have different effects on
developmental processes when they occur at different points in
time. Because individuals with low DI require more powerful
doses of specific factors to develop a disorder, unique factors
may be best identified in individuals relatively low in DI.
We distinguish DI from those factors specific to particular
disorders, but recognize that origins of phenotypic atypicality lie
along a continuum. Some reflect predominantly generalized
factors and others relatively unique factors. Nonetheless, the
key components of the DI model hold: Causal factors that have
relatively generalized effects exert them through DI, and these
interact with both specific genetic and environmental effects to
give rise to atypical neurodevelopmental outcomes.
The DI model offers several testable predictions. First, an
increased incidence of the markers of DI should be found in each
relevant neurodevelopmental disorder. Second, we should
be able to observe a greater incidence of known correlates of
DI, such as atypical functional and anatomic brain symmetry
and reduced intellectual functioning, in each disorder. Third,
because DI increases the risk of atypical brain development in
general, high-DI individuals may well show features of several
different neurodevelopmental disorders—in other words,
comorbidity. Below, we summarize empirical support for the
model in three specific neurodevelopmental disorders.
EMPIRICAL SUPPORT: ATTENTION-DEFICIT/
HYPERACTIVITY DISORDER (ADHD), DYSLEXIA, AND
SCHIZOPHRENIA
Markers of DI
Several studies have observed that children with hyperactivity
and dyslexia have an elevated incidence of MPAs and there is an
extensive literature documenting increased MPAs in schizo-
phrenia (Yeo et al., 1999). Fewer studies have examined FA.
Two studies have examined relations between FA and ADHD
symptoms in normal college populations (Stevenson et al.,
2006). In the first, greater total FA predicted more ADHD
symptoms in males. In the second, greater body FA predicted
greater symptoms in women only; the different pattern of results
across studies was attributed to sampling variability. Future
work should examine whether these effects are stronger in
individuals carrying the full diagnosis of ADHD. We investi-
gated the incidence of FA and MPAs in dyslexia and schizo-
phrenia (Edgar, Yeo, Canive, & Miller, 2006). A composite index
was greater in both dyslexics and schizophrenics than in
controls. Considered individually, the MPA score differed
from controls in both groups, although the FA measure did not.
247
 Developmental Instability and Neurodevelopmental Disorders
Several studies have found greater FA in people with schizo-
phrenia (Yeo et al., 1999). Schizotypy (a set of personality traits
found in schizophrenia) is also associated with greater FA (Rosa
et al., 2000).
Correlates of DI
In ADHD, a variety of atypical asymmetries have been reported,
including atypical hand, foot, ear, and eye dominance; atypical
linguistic- and emotional-processing asymmetries; and an ele-
vated incidence of atypical asymmetry of the planum temporale
(an area of the temporal lobe that is associated with language
processing). Atypical handedness, language lateralization, and
asymmetry of the planum temporale are commonly observed in
schizophrenia (Edgar et al., 2006; Yeo et al., 1999). We recently
noted atypical lateralization of evoked brain activity in both
schizophrenics and dyslexics using magnetoencephalography
(Edgar et al., 2006).
Modest reductions in general intellectual functioning are
characteristic of children and adults with ADHD, and the entire
association between ADHD and IQ can be accounted for by
shared genetic influences (Kuntsi et al., 2004). Reduced verbal
IQ, and to a lesser extent performance IQ, are typically noted in
people with dyslexia. Relatively lower intellectual functioning is
also a risk factor for schizophrenia, and it represents a vulner-
ability factor for, rather than a consequence of, the emerging
disease (e.g., Reichenberg et al., 2006). One possible explana-
tion worthy of further investigation is that reduced intelligence
in these disordered populations reflects increased DI.
Genetics and Comorbidity
ADHD, dyslexia, and schizophrenia are each heritable and in-
fluenced by many genes. Individual genes identified to date have
relatively small effects. Comorbid conditions are common. A
large-scale study of psychiatric comorbidity among adults with
ADHD found elevated rates of mood disorders, anxiety disor-
ders, and substance abuse problems (Kessler et al., 2006). An-
other large-scale study of children diagnosed with ADHD found
that 50% also had a learning disability and 17% also had a
behavioral disorder (Decker, McIntosh, Kelly, Nicholls, & Dean,
2001). People with dyslexia tend to have elevated levels of
schizotypy, and an excess of schizophrenia is found in people
with dyslexia and their relatives (Yeo et al., 1999). Specific ge-
netic loci for dyslexia overlap with those for IQ (Posthuma et al.,
2005). Twin pairs in which both individuals have schizophrenia
have greater FA than twin pairs in which only one individual has
schizophrenia. Greater paternal age, which is associated with
greater mutation load, has been linked with both schizophrenia
and reduced intelligence (Malaspina et al., 2005). Again, further
investigation can address the role of DI in generating these
associations.
The DI perspective may also provide insight into one of the
most vexing problems in psychiatric genetics: how the genes for
248
schizophrenia are maintained despite abundant evidence that
schizophrenics, especially males, have fewer children and
hence pass on fewer of their genes to the next generation. The
existence of ‘‘schizophrenia genes’’ largely reflects the continual
operation of the engines of genetic diversity, including the
formation of new mutations and genetic conflict. This perspec-
tive is consistent with one large three-generation study, which
found no compensating fertility advantage to relatives of
schizophrenic individuals (Svensson et al., 2007).
CONCLUSIONS AND FUTURE DIRECTIONS
The evolutionary genetics of mental illness and learning
disabilities cannot be disentangled from the evolutionary genetics
of normal variation. Understanding how individually advanta-
geous or disadvantageous genes aggregate in individuals or
families requires an understanding of human mate selection.
Our theory posits two interacting causal factors—those as-
sociated with DI and those that are disorder specific—and
produces the following questions for further investigation:
 Do causal factors interact in different ways for different
disorders? One might expect that more common disorders
require fewer additional specific factors and should thus
more directly result from DI.
 Does the developmental timing of DI matter? Different
measures of DI (MPAs, FA of skeletal features, FA of dermal
ridges on the palms and fingers), which reflect perturbation
during different time windows, may inform us of the critical
developmental periods for different disorders.
 Can measures of DI be used as tools to clarify the environ-
mental and genetic causes of specific disorders? The study
of affected individuals relatively low in DI may be most re-
vealing of specific genetic influences or brain abnormalities
(because such individuals require a bigger ‘‘dose’’ of genetic
influences or brain abnormalities to exhibit disorders), while
study of individuals high in DI may be most revealing of
environmental stressors (because those individuals are more
sensitive to such stressors).
The DI model offers much promise. Although the amount of re-
search incorporating measures of DI has steadily increased in the
past decade, much more empirical work and theoretical refinement
is needed. We urge researchers to, when feasible, regularly include
measures of DI in their studies of neurodevelopmental disorders.
Recommended Reading
Kowner, R. (2001). Psychological perspective on human developmental
stability and fluctuating asymmetry: Sources, applications, and
implications. British Journal of Psychology, 92, 447–469.
Polak, M. (Ed.), (2003). Developmental instability: Causes and conse-
quences. Cambridge, UK: Cambridge University Press.
Volume 16—Number 5
 Ronald A. Yeo, Steven W. Gangestad, and Robert J. Thoma
REFERENCES
Decker, S.L., McIntosh, D.E., Kelly, A.M., Nicholls, S.K., & Dean, R.S.
(2001). Comorbidity among individuals classified with attention
disorders. International Journal of Neuroscience, 110, 43–54.
Edgar, C., Yeo, R.A., Canive, J., & Miller, G. (2006). Reduced auditory
M100 asymmetry in schizophrenia and dyslexia: Applying a de-
velopmental instability approach to assess atypical brain asym-
metry. Neuropsychologia, 44, 289–299.
Gangestad, S.W., Thornhill, R., & Garver-Apgar, C.E. (2005). Adap-
tations to ovulation: Implications for sexual and social behavior.
Current Directions in Psychological Science, 14, 312–316.
Kessler, R.C., Adler, L., Barkley, R., Biederman, J., Connors, C.K.,
Demler, O., et al. (2006). The prevalence and correlates of adult
ADHD in the United States: Results from the national comorbidity
survey replication. American Journal of Psychiatry, 163, 716–723.
Kuntsi, J., Eley, T.C., Taylor, A., Hughes, C., Asherson, P., Caspi, A., &
Moffitt, T.E. (2004). Co-occurrence of ADHD and low IQ has
genetic origins. American Journal of Medical Genetics Part B
(Neuropsychiatric Genetics), 124B, 41–47.
Malaspina, D., Reichenberg, A., Weiser, M., Fennig, S., Davidson,
M., Harlap, S., et al. (2005). Paternal age and intelligence: Im-
plications for age-related genomic changes in male germ cells.
Psychiatric Genetics, 15, 117–125.
Mller, A.P. (2006). A review of developmental instability, parasitism
and disease infection, genetics, and evolution. Infection, Genetics,
and Evolution, 6, 133–140.
Posthuma, D., Luciano, M., de Geus, E.J.C., Wright, M.J., Slagboom,
P.E., Montgomery, G.W., et al. (2005). A genomewide scan
for intelligence identifies quantitative trait loci on 2q and 6p.
American Journal of Human Genetics, 77, 318–326.
Volume 16—Number 5
Prokosch, M.D., Yeo, R.A., & 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, 203–213.
Reichenberg, A., Weiser, M., Rapp, M.A., Rabinowitz, J., Caspi,
A., Schmeidler, J., et al. (2006). Premorbid intra-individual
variability in intellectual performance and risk for schizophre-
nia: A population-based study. Schizophrenia Research, 85,
49–57.
Rosa, A., van Os, J., Fananasd, L., Barrantes, N., Caparros, B., Gui-
terrez, B., & Obiols, J. (2000). Developmental instability and
schizotypy. Schizophrenia Research, 43, 125–134.
Rose, R.J. (2005). Prenatal programming of behavior: A twin-study
perspective. Neuroscience and Biobehavioral Reviews, 29, 321–
327.
Stevenson, J.C., Everson, P.M., Williams, D.C., Hipskind, G., Mahoney,
E.R., Mehler, M., et al. (2006). Attention-deficit hyperactivity
disorder and fluctuating asymmetry in another college sample.
American Journal of Human Biology, 18, 402–414.
Svensson, A.C., Lichtenstein, O., Sanden, S., & Hultman, C.M. (2007).
Fertility of first-degree relatives of patients with schizophrenia:
A three-generation perspective. Schizophrenia Research, 91, 238–
245.
Yeo, R.A., Gangestad, S.W., Edgar, C., & Thoma, R. (1999).
The evolutionary-genetic underpinnings of schizophrenia: The
developmental instability model. Schizophrenia Research, 39,
197–206.
Yeo, R.A., Thoma, R., & Gangestad, S.W. (2002). Human handedness:
A biological perspective. In I. Rapin & S. Segalowitz (Eds.),
Handbook of neuropsychology (pp. 329–364). Amsterdam:
Elsevier Science.
249