Chapter 14
Heritability and Socially
Defined Race Differences
in Intelligence
Key themes
Genetic heritability of intelligence
Genetic and environmental influences on intelligence
The bell curve
Socially defined race differences in intelligence
Eugenics in intelligence research
Learning outcomes
By the end of this chapter you should:
Understand what is meant by genetic heritability in intelligence
Be able to outline the different dimensions of genetic and
environmental influences that are thought to affect intelligence
Be familiar with the main points of Herrnstein and Murray’s bell
curve analysis
Be aware of some criticisms of Herrnstein and Murray’s bell
curve analysis
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 336
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 337
In 1994 Arthur Jensen, a prominent IQ psychologist,
wrote:
Consideration of the book’s actual content is being
displaced by the rhetoric of denial: name calling
(‘neo-Nazi’, ‘pseudo-scientific,’ ‘racism’), sidetracks
(‘but does IQ really measure intelligence?’),
non sequiturs (‘specific genes for IQ have not been
identified, so we can claim nothing about its heritability’),
red herrings (‘Hitler misused genetics’),
falsehoods (‘all the tests are biased’), hyperbole
(‘throwing gasoline on a fire’), and insults (‘creepy’,
‘indecent’, ‘ugly’).
(Jensen, 1994, p. 48)
What book created these descriptions, and why did
Jensen feel it necessary to defend such a book? In this
chapter we address the history, theory and debate that
surround the examination of race differences in intelligence,
and explore the debates and consequences of
some of the conclusions that are drawn from this
research. It is not a pretty side of psychology. Assertions
and evidence presented are not always palatable to
many. Group differences in intelligence are an interesting
area of debate, but it is not without its dark side. And
as part of this chapter, we are going to explore the
actual content of the aforementioned book.
This chapter is slightly longer than many others in the
book. In this chapter we deal with some controversial
and sensitive issues in the psychology of intelligence.
We argue that many areas need to be explained and
fully explored. Additionally, we introduce a number of
concepts that you have come across in previous chapters,
so rather than suggest you spend your time going
backwards and forwards through the book, we have
spent a little time restating the main issues and findings
from other discussions, to make it clear how these issues
apply to the debates in this chapter. Consequently,
you should find some of this material easier going
because you have been introduced to a lot of the
arguments in previous chapters.
We realise that there is slightly more material here, so
we have split the chapter into two main sections. Please
remember, however, that these two areas are intrinsically
linked.
Section A – The heritability of intelligence – In this
section we will outline what has become known as
the nature versus nurture debate, in which we
Introduction
compare and consider genetic versus environmental
effects on intelligence. Key themes in this debate
include heritability of intelligence and genetic and
environmental influences on intelligence.
Section B – The bell curve: race differences in
intelligence – In this section we will outline what has
become known as the bell curve debate. We consider
evidence and arguments regarding race differences
in intelligence. Key themes in this debate
include an outline and consideration of a bell curve,
group differences, most notably race differences in
intelligence, and the role of eugenics in intelligence
research.
Source: Getty Images/Tim Platt
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 337
338 PART 2 INTELLIGENCE
Section A – The heritability
of intelligence
Intelligence: the nature versus
nurture debate
Here we will outline what has become known as the nature
versus nurture debate.We will compare and consider genetic
versus environmental effects on intelligence.
Galton
In 1865, Sir Francis Galton began to study the heritability
of intelligence, following his reading of his cousin
Charles Darwin’s publication The Origin of Species,
which dealt with the idea that all species gradually evolve
through the process of natural selection. In following
this work, Galton soon became interested in studying the
variations in human ability, and particularly intelligence.
In his book Hereditary Genius (Galton, 1869), he began
investigating why higher intelligence seemed to run in
families. He suggested that man’s natural abilities are
inherited under the same conditions as physical features
of the animal world described by Darwin. Galton suggested
that children inherit their intelligence from their
parents.
To support such an assertion, Galton started analysing
the obituaries of The Times newspaper so that he could
identify the ancestry of eminent men. What Galton did
was to compare different degrees of relationship between
individuals in terms of being biological relatives (i.e., parents,
siblings, cousins) and the eminence of each of these
individuals. First-degree relatives are relatives with whom
an individual shares an estimated average of 50 per cent
(half) of their genes (though note this is an estimated average
percentage; you will learn more about why this is an
estimated average in this chapter). First-degree relatives
include your parents, brothers and sisters and children. A
second-degree relative is a relative with whom an individual
shares an estimated average of 25 per cent (a quarter)
of their genes, that is, grandparents, grandchildren, aunts,
uncles, nephews, nieces. A third-degree relative is a relative
with whom an individual shares an estimated average
of 12.5 per cent (one-eighth) of their genes. Third-degree
relatives include your great-grandparents, great-aunts,
great-uncles and first cousins. Galton found that the number
of eminent relatives of an eminent person was greater
for first-degree relatives than for second-degree relatives;
and again, the number was greater for second-degree relatives
than for third-degree relatives. This result suggested
to Galton that there is evidence for the heritability of
intelligence.
However, Galton quickly became concerned with
whether intelligence was simply heritable, or whether it
was also influenced by the environment. It was here that
Galton was the first psychologist to make the distinction
between ‘nature’ and ‘nurture’ (and he was the first to use
this now-common phrase). To examine this idea, he surveyed
190 Fellows of the Royal Society, of which Galton
was a member (Galton, 1874). The Royal Society is a highly
prestigious scientific society dedicated to establishing the
truth of scientific matters through experiment. It has had
several famous scientists as members, including Robert
Boyle, Sir Christopher Wren and Isaac Newton. Galton
asked his fellow members of the Society several questions
regarding their birth order and the occupation and race of
their parents. He wanted to find out whether members of
the Society’s achievements and interest in science were due
to their natural makeup (nature) or to their environment,
for example, the encouragement of their talents by others
(nurture).
You have to remember that many of Galton’s speculations
arose before we knew as much about genetics as we
do today. That is why many recognise him as a truly great
scientist. Galton himself recognised the inherent problems
of such studies (Galton, 1875). For example, he
speculated about the confounding effects of the environment
and realised that eminent people might not have
arisen to their current status alone, but with the help of
relatives. Galton believed that the question of whether nature
or nurture influences intelligence could be examined
more carefully by comparing twins. He suggested that
comparisons of twins who were similar at birth but had
grown up in different environments, and comparisons of
dissimilar twins who had grown up in similar environments,
might hold the key to examining the nature–nurture
debate surrounding intelligence. He also proposed that
adoption studies might be useful to analyse the different
effects of heredity and environment. His speculations
about twins and adoption studies laid the groundwork for
modern attempts to examine the nature-nurture debate
in intelligence.
Heritability of intelligence
Within the nature versus nurture consideration of intelligence,
we find ourselves concentrating on behavioural genetic
principles (you will have come across many of these
ideas if you have read Chapter 8). One area of behavioural
genetics concentrates on the relationships between genes
and environment, to compare the similarities and differences
between individuals within a particular population
and assess the relative influence of genes and the environment
on any behaviour. In this case, the behaviour we are
looking at is intelligence.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 338
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 339
What do we mean by heritability
of intelligence?
Behavioural geneticists such as Robert Plomin have written
extensively about behavioural genetics (Plomin, 2004;
Plomin, DeFries, McClearn and McGuffin, 2000). Heritability
of intelligence begins with the fact that genes are
biologically transmitted from biological parents to the
child. Children inherit 50 per cent of their father’s genes
and 50 per cent of their mother’s genes. We can use this
information as a starting point to explore how genes
influence intelligence.
The assessment of the extent to which any phenotype
(any outward manifestation of the individual – physical attractiveness,
behaviour, intelligence) is passed on from parents
to children, from the results of their genes, is termed as
genetic heritability. The genetic heritability of any phenotype
is assessed in terms of variability (i.e., how much
they differ) between the parents and the child. This variability
is often assessed within the proportion of shared
variance of that behaviour between the parent and child.
Proportion of shared variance is presented as a percentage
(i.e., out of 100 per cent). When a parent and child are very
similar in a particular characteristic, there is thought to be
a low variability between parent and child, and the proportion
of shared variance of that behaviour is high (nearer
100 per cent). In other words, the parent and child are not
very different in this characteristic. Conversely, when a parent
and child are quite different in a particular characteristic,
there is thought to be a high variability between parent
and child, and the proportion of shared variance of that
behaviour is low (nearer 0 per cent).
The heritability of a human physical characteristic, such
as having a nose, is entirely genetic and not in any way
influenced by factors such as the environment. In fact the
environment is seen as having zero variability, or a proportion
of shared variance of 100 per cent. However, some
aspects of human behaviour (including intelligence), in
which the environment is thought to have an influence,
have greater amounts of heritable variability and lower
shared variance. For example, choosing which football
team to support would be heavily determined by environmental
factors such as where you are born, your parents’
football team, your friends, and the first football team you
see. Choosing a favourite football team has high variability
between parent and child, but the proportion of shared
variance of favourite football team due to genetic heritability
would be zero (0 per cent).
In behavioural genetics of intelligence, researchers are
primarily interested in estimating the extent of genetic heritability
of intelligence across a population, and stating the
genetic heritability of that behaviour in terms of shared
variance. This estimated average of genetic heritability is
known as h2
. Therefore, h2
is the average estimate of the
proportion of variance for intelligence thought to be accounted
for by genetic factors across a population.
You may have noticed we emphasised estimating,
estimate (‘estimate’ meaning to calculate approximately)
Galton felt there was some value in studying eminent families as an indicator of
the heritability of intelligence. How well would such an approach work today?
Source: Stuart Atkins/Rex Features
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 339
and average. This is because, for a long time in psychology,
for any phenotype (characteristic or behaviour) the estimates
of the strength of genetics factors were done and interpreted
within a process called the additive assumption.
This additive assumption suggests that there are only two
dimensions that determine heritability of any behaviour
(in our case, intelligence): (1) the genetic part, which we’ve
just outlined, and (2) the environment. Consequently,
overall, heritability of intelligence is estimated in terms of
the relative strength of both (e.g., nature versus nurture).
Therefore, the influence of genetic (G) and environmental
(E) components, in this theory, will always add together to
account for 100 per cent of the variance of intelligence. On
the basis of this assumption, the heritability coefficient (h2
)
can be subtracted from 100 per cent to calculate the environmental
contribution to intelligence. If researchers computed,
for example, that genetics accounted for an average
of 25 per cent of the variance for intelligence, we would estimate
that the environmental factors account for an average
of 75 per cent of the variance of intelligence. However,
it is important to note that the additive assumption is now
considered a starting point for calculating heritability of
intelligence and for estimating the amount of genes that
people are expected to share (e.g., brothers and sisters are
expected to share 50 per cent). We will see later in this
chapter that this view of assessing heritability has changed
a lot. The idea of determining the relative strength of
genetics and environmental factors by simply adding together
genetic and environmental factors is more complicated
than once thought, and psychologists really do
emphasise the words‘estimate’ and‘average’ when referring
to heritability.
Methods for assessing genetic
heritability of intelligence
So, how might we assess genetic influences on intelligence?
Well, as Galton himself mentioned, the relationship between
genes and intelligence has traditionally been studied
by concentrating on the similarities and differences between
populations of individuals to assess the relative
influence of their shared genes in intelligence.
Plomin (2004) identifies three main types of studies
that use this technique: family studies, twin studies and
adoption studies. As children share an estimated average of
50 per cent of their genes with each of their parents, and
they also share genes with their brothers and sisters, it is of
interest to behavioural genetics researchers to examine
possible associations between parents’ and children’s behaviours
within a family. This leads to the first type of
study, family studies. However, these studies on their own
potentially tell us very little because all children share an
estimated average of 50 per cent of their genes with each of
their parents and with their brothers and sisters. As well as
this, using observation, interview or questionnaire measures
also presents a problem because similarities between
personalities might be due to environmental influence (i.e.,
an intelligent daughter might be like her extraverted
mother because she copies her behaviour). These are real
concerns until we consider the occasions when families
don’t typically share genes in this way. There are two main
examples: twin studies and adoption studies.
Twin studies provide an interesting area of research, as
there is a possibility of comparing different types of genetic
makeup so as to compare genetic influences. The term
‘twin’ refers to two individuals who have shared the same
uterus (the uterus or womb is the major female reproductive
organ). Identical (or monozygotic) twins occur when a
single egg is fertilised to form one zygote, but the zygote
then divides into two separate embryos. The two embryos
develop into fetuses sharing the same womb. Identical
twins are always of the same sex and have the same
arrangement of genes and chromosomes (which contain
the hereditary information necessary for cell life). These
twins share 100 per cent of genes with each other. Fraternal
twins (non-identical twins, or dizygotic) usually occur
when two fertilised eggs are implanted in the uterine wall
at the same time. The two eggs form two zygotes (hence
they are dizygotic). These twins share an estimated average
of 50 per cent of their genetic makeup. Consequently, some
researchers compare behaviours across non-twins, identical
and fraternal twins to examine the relative influence of
genetics.
The influence of the environment and genetics is often
compared in adoption studies. Intelligence can be compared
between parents and adopted children, as there is no
genetic heritability. Variables are often compared between
siblings, or twins, reared apart to examine the extent of genetic
and environmental effects. For example, if two twins
show similar behaviours, despite being raised in different
environments, this suggests that genes may be important in
that behaviour.
Once you consider all these types of studies together, in
which intelligence is compared between parents and children,
and siblings that share 0–100 per cent genetic similarity,
you can begin to make assessments of the extent of
genetic heritability across a population.
It is important to remember that there is no physiological
procedure in these sorts of studies. Behaviour geneticists
do not have the ability to assess the genetic heritability
of intelligence using advanced biological measures, or a
complex scientific genetic analysis (well, not yet). Rather,
researchers look for similarities and differences in intelligence
among individual people by using observation,
interview or questionnaire measures. They look for similarities
between parents’ and children’s intelligence (using
intelligence measures) to determine the extent of genetic
340 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 340
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 341
influence on intelligence. What is also important to remember
is that, when we deal with heritability estimates,
we don’t talk about heritability estimates for particular individuals;
rather, researchers estimate the average heritability
among certain populations of people – monozygotic
(MZ, identical) twins, dizygotic (DZ, fraternal) twins, family
members, parents and children. So an heritability estimate
of 50 per cent for intelligence does not mean that we
all inherit 50 per cent of that intelligence trait from our
genes; it means that across the population, the genetic heritability
of intelligence has been estimated at an average of
50 per cent.
Heritability estimates of intelligence
What is the heritability of intelligence from these types of
studies? Well, some studies have estimated the heritability
of intelligence based on family, twin and adoption studies.
For example, there have been a number of findings from
Bouchard’s Minnesota Study of Twins Reared Apart (overseen
by US behavioural geneticist Thomas Bouchard). This
research involves not only the medical and psychological
assessment of identical (monozygotic) and fraternal (dizygotic)
twins separated early in life and reared apart, on
which figures are given, but also their intelligence. A wellcited
documentation of these studies was recently provided
by behavioural geneticist journalist Matt Ridley (Ridley,
1999). Ridley put together all the modern family, twin and
adoption studies, which mainly included the findings of
Bouchard and McGue’s meta-analysis of 111 studies
(Bouchard and McGue, 1981). The following analysis by
Ridley is the concordance rate of IQ (the presence of the
same intelligence level between two individuals) from all
these studies (in parentheses are concordance rates given by
Bouchard and McGue’s meta-analysis; see also Figure 14.1):
100%: Perfect concordance rate;
87%: Same person tested twice;
86%: Identical twins reared together (86%);
76%: Identical twins reared apart (72%);
55%: Fraternal twins reared together (60%);
47%: Biological siblings reared together (47%);
40%: Parents and children living together (42%);
31%: Parents and children living apart (22%);
24%: Biological siblings reared apart (24%);
15%: Cousins (Bouchard and McGue only);
00%: Adopted children living together;
00%: Unrelated people living apart.
Evidence such as this, coming from studies throughout
the century, was consistent, and researchers were able to
make estimates on the level of genetic heritability of intelligence.
You can see from evidence like this how people
Person
taking the
same IQ
test twice
Identical
twins
reared
together
Identical
twins
reared
apart
Fraternal
twins
reared
together
Biological
siblings
reared
together
Biological
siblings
reared
apart
Adopted
children
living
together
(0%)
Unrelated
people
living
apart
(0%)
Parents
and
children
living
together
Parents
and
children
living
apart
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
87%
24%31%
40%
47%
55%
76%
86%
Figure 14.1 Concordance rates of intelligence.
Source: From Ridley, M. (1999). Genome: the autobiography of a species in 23 chapters. London: Fourth Estate. Reprinted by permission of
HarperCollins Publishers Ltd. © M. Ridley 1999.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 341
would tend to estimate the influence of genetics on intelligence
as in some instances relatively high, because the evidence
for heritability, in some instances, is over 80 per cent.
For example, Professor Hans Eysenck (Eysenck, 1979) used
this sort of evidence to suggest that the estimation of heritability
of intelligence was around 69 per cent in the general
population. Later, Herrnstein and Murray (1994),
whose work we will discuss at greater length later, estimated
heritability in the general population at 74 per cent.
However, we know a lot more about the influences of
genetics on intelligence today. We outlined some of these
influences in Chapter 8, but we are going to outline them
fully here so you can see their relevance to the literature on
intelligence as well as how the estimates of the genetic
influence on intelligence might be lower than previously
estimated.
Considerations within behavioural
genetics and intelligence
The idea of how genes and the environment are viewed,
and used, to predict the heritability of intelligence has
changed over recent years.
Authors such as US psychologist E. E. Maccoby (2000)
and Robert Plomin (2004) suggest the additive principle of
determining heritability of intelligence (or any phenotype)
is not applicable any more. The validity of the additive assumption
in computing the relative strength of genetics
and environment in determining behaviour has been
widely challenged. The first problem is that estimating the
environment is usually done without utilising any direct
measures of environmental factors. For example, researchers
often compute genetic heritability and then subtract
that from 100 per cent. Obviously, if the estimates of
heritability are indeterminate, or prone to error, so are the
estimates of E derived by subtracting heritability from
100 per cent. A further problem with the additive assumption
of computing heritability is that when genetic heritability
is large, it assumes that all environmental factors
associated with that behaviour must be small. Therefore, it
is better to see human intelligence as a joint result of an
interaction between a person’s genes and environmental
factors. Intelligence should not be seen as the result of
‘genetics ϩ environment’ but rather of ‘genetics ϫ environment’.
For example, it is better to view the relative influences
of genes and environment on intelligence as the
result of a long-term interaction, with environmental factors
triggering certain genetic behaviours and the effects of
the environment differing between individuals because of
their genetic makeup.
What is important for you to note is that these changes
and developments in research and thinking have been suggested,
encouraged and developed by both theorists and researchers,
many of whom we have already mentioned, who
support and criticise the idea of genetic heritability in intelligence.
So, what has brought about, and resulted from,
such a general shift in thinking, from the additive principle
of ‘genetics ϩ environment’ to the later, more integrative
idea of ‘genetics ϫ environment’? Well, four considerations
surrounding modern-day thinking in behavioural genetics
are important when considering any phenotype, particularly
intelligence (see Figure 14.2):
conceptions of heritability and the environment;
different types of genetic variance;
the representativeness of twin and adoption studies;
assortative mating.
342 PART 2 INTELLIGENCE
Considerations
to be made
regarding genetic
heritability
Conceptions of
heritability and
the environment
Assortative mating
The representativeness
of twin and adoption
studies
Different types of
genetic variance
Figure 14.2 Considerations to be made regarding genetic heritability.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 342
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 343
Conceptions of genetic heritability
and the environment
Gregory Carey (2002) suggests that there are two important
contexts within which to consider heritability and
environmental influences on intelligence:
Abstract concepts – These are generally theoretical
(not applied or practical) concepts. As Carey explains,
whatever the numerical estimates of either genetic or
environmental influences, they provide us with little information
about the specific genes, or specific environmental
variables, that influence intelligence.
Population concepts – All of these estimates refer to
any group of people considered as a population, but
they tell us very little about any single individual. For
example, just because intelligence may have a genetic
heritability of around 60 per cent, it does not mean, for
any one individual, that 60 per cent of their intelligence
is due to genes and 40 per cent of their intelligence is
due to the environment. Rather, it is estimated across
the population that genetic heritability of intelligence is
at an average of around 40 per cent, and individuals will
vary around that estimate.
Different types of genetic variance
So far, in this chapter, we have treated genetic influence on
intelligence only as a single entity, namely, the influence of
your genes on your intelligence. However, behavioural
genetics researchers such as Thomas Bouchard and
M. McGue (Bouchard and McGue, 1981) note that genetic
influence does not simply comprise one aspect, but in fact
three aspects: (1) additive genetic variance, (2) dominant
genetic variance and (3) epistatic genetic variance.
Additive genetic variation is the genetic variance that
we have previously described in this chapter, which is
genetic variation in behaviour that is the total of the
individual’s genes inherited from their parents.
However, the two other types of genetic variation are
known as non-additive genetic variance.
First, dominant genetic variance is part of a process
by which certain genes are expressed (dominant genes)
and other genes are not expressed (recessive genes). Every
one of us has two copies of every gene, one inherited from
our mother and one from our father. Sometimes the two
genes, which determine a particular trait (for example,
eye colour) will actually code for two types of characteristics
(for example, blue eyes and brown eyes). If one of
these genes is dominant, then only its character is expressed
and not that of the other gene. For example, if
blue eyes were a dominant gene, and your mother had
brown eyes and your father had blue eyes, you would inherit
blue eyes.
Secondly, epistatic genetic variance (known as interactive
genetic variance) refers to a process by which
genes interact. It is now known that several different genes
not only influence physical characteristics and behaviour
on their own, but work and interact together. Unlike dominant
genetic variance, which just applies to one gene replacing
another, epistatic genetic variance is the result of
the way certain genes that we inherit determine whether
other genes we inherit will be expressed or suppressed (this
process is called epistasis).
It is difficult to measure dominant genetic variance and
epistatic genetic variance when it comes to intelligence.
However, it is now accepted that these three aspects – additive
genetic variance, dominance genetic variance and
epistatic genetic variance – are thought to make up total
genetic variance of intelligence.
You can see the genetic side of things is a lot more complicated
than just viewing genes as a single entity, as genes
interact and suppress other genes. You will see, in the literature,
behavioural geneticists referring to terms such as ‘narrow
heritability’ and ‘broad heritability’. Narrow heritability
is just additive genetic variance. Broad heritability is all three
aspects of genetic heritability (additive genetic variance ϩ
dominant genetic variance ϩ epistatic genetic variance).
Consequently, authors such as Thomas Bouchard and
M. McGue (1981) and US psychologists Heather Chipeur,
Michael Rovine and Robert Plomin (1990) have suggested
that original estimates of the average percentage of
parental genes that children inherit, and siblings share, may
have been previously oversimplified. For example, these
authors suggest that genetic variations in heritability of any
phenotype (including intelligence) should be made in the
following terms:
identical (monozygotic; MZ) twins ϭ additive genetic
variance ϩ non-additive genetic variance (where previously
it was presumed to be just additive genetic vari-
ance);
fraternal (dizygotic; DZ) twins ϭ 0.5 of additive genetic
variance ϩ 0.25 of non-additive genetic variance (rather
than just 0.5 of additive genetic variance).
As you can see, computing the level of genetic variance
may be more complicated than previously thought. Today,
behavioural geneticists take these factors into account
when suggesting the strength of heritability estimates of
intelligence.
Problems with the representativeness
of twin and adoption studies
One of the considerations put forward by psychologists
such as Maccoby and by Leon Kamin and Arthur
Goldberger (2002) concerns adoption and twin studies. A
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 343
significant portion of studies examining heritability effects
is devoted to twin and adoption studies. Twin studies are
important because they allow the comparison of different
types of twins to compare genetic influences; monozygotic
(MZ, identical) twins who share 100 per cent of their
genes, and dizygotic (DZ, fraternal) twins who share 50 per
cent of their genes. Adoption studies are important because
they include two sets of factors that may account for differences
in behaviour: biological parents and environmental
parents. Therefore, it is argued that these families aren’t
necessarily representative of the normal populations. This
natural bias in sampling may underestimate or overestimate
heritability estimates across the general population
because genetic influences in these samples may not be representative
of the whole population.
This issue is particularly important when considering
research that assesses heritability of intelligence using twin
and adoption studies. Leon Kamin and Arthur Goldberger
(2002) estimate that twin studies might overestimate the
role of genetics, particularly because identical twins have
more similar environments than do same-sex fraternal
twins. Also, research shows identical twins are treated more
similarly by their parents, spend more time together and
more often have the same friends. Their environmental experience
comprises a greater proportion of each other’s
social environment than does that of fraternal siblings.
Consequently, if genetic heritability estimates are usually
larger in twin studies than in adoption studies, then some
of the estimated similarity that is attributed to genetic influence
might not be correct. Stoolmiller (1998) has also
suggested that adoption studies lead to a similar restriction
of the measurement of environmental factors. Stoolmiller
argues that the placement strategies of adoption agencies
might influence heritability estimates. For example, adoption
agencies might always place children in affluent or
middle- to high-income families; thus the effects of economic
status are never fully explored in these studies,
because an adopted child would rarely be placed into a
household suffering from poverty.
Assortative mating
Nicholas Mackintosh (Mackintosh, 1998), animal-learning
theorist at the University of Cambridge, raises the
issue that assortative mating can have an effect on genetic
variance and, therefore, on estimates of genetic heritability
of intelligence. Assortative mating is a complicated
name for the simple concept that, when couples mate, they
either have several traits in common or contrast wildly in
their traits. A lot of the understanding of genetic variation
is based on the assumption that two individuals mate
quite randomly with random people and, therefore, that
any genetic similarity between them is by chance. But we
know this is not true. We know that people mate with people
who they perceive are similar to themselves. For example,
we tend to see people mating with people who are of a
similar size, or similar in their ‘good-lookingness’. This is
called positive assortative mating. Equally, we find people
mating who are completely the opposite. This is called
negative assortative mating. Therefore, in much the same
way, the assortative mating principle can be applied to
intelligence. That is, individuals may seek to mate with
people who are of a similar intelligence. Think about your
parents; do they have a similar educational background to
each other?
Modern estimates of the genetic
heritability of intelligence
In the light of the considerations just outlined, different researchers
have tried to estimate the overall heritability of
intelligence across the general population. These estimates
tend to be broad estimates of genetic variance (additive
and non-additive genetic variance) in intelligence. They
break down by what percentage of intelligence is determined
by genetics and what percentage can be attributed to
the environment. As we have already mentioned, higher estimates
of heritability have come from authors like Hans
Eysenck (1971, 1991), who once estimated it at 69 per cent.
There is no evidence to suggest that this is wrong; however,
modern-day commentators, given the preceding issues relating
to heritability estimates, are more conservative; that
suggests heritability falls into a range. The American Psychological
Association task force (that we mentioned in
Chapter 10), headed by Ulric Neisser, estimates the heritability
of intelligence ranges from 40 to 80 per cent
(Neisser et al., 1996), while Nicolas Mackintosh (1998) suggests
a range of 30 to 75 per cent. If you are looking for a
more exact figure, Chipeur et al. (1990) have suggested the
genetic heritability of intelligence at 50 per cent which is a
commonly – but not always – accepted viewpoint.
All this discussion, and we haven’t even considered the
extent of environmental effects on intelligence yet!
Environmental influences
on intelligence
The list of possible environmental effects on intelligence
could be endless. It could range from the effects on intelligence
of long-term poverty (for example, never getting the
opportunities to develop skills at home, at school and consequently
in a career), through to one conversation with one
teacher who suggests that the person will never amount to
much.Thomas Bouchard and Nancy Segal (1985) list 21 factors
that are related to intelligence, including malnutrition,
344 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 344
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 345
Stop and think
Cyril L. Burt
For many lecturers and students across the psychology
discipline, Cyril Burt is psychology’s equivalent of the
bogeyman. Like parents who tell children who are
naughty that the bogeyman will come and get them, lecturers
inform students who are naughty and make up
their data that they will be branded with the Cyril Burt
label. No one knows the truth about Cyril Burt; but stories
of acts done, a long time ago, are told through generations
of lecturers. Your lecturers were told by their
lecturers about Cyril Burt, and though many stories
about him may be untrue, academics today still argue
and swap conspiracy, and counterconspiracy, stories
surrounding Cyril Burt.
Cyril L. Burt was born on 3 March 1893 and died
10 October 1971. He was a British educational psychologist.
He studied at Oxford in the United Kingdom, and
in Germany. Between 1908 and 1971, he worked at the
University of Liverpool and had a chair of Psychology at
University College, London. He also worked as Chief
Psychologist at London County Council. He was President
of the British Psychological Society (in 1942), was
editor and co-editor of the British Journal of Statistical
Psychology (between 1947 and 1963) and was the first
psychologist to be knighted (in 1946). He founded the
field of Educational Psychology in the United Kingdom
and helped to establish Eleven-Plus testing in Great
Britain.
In his academic work, Burt published nine books
and more than 300 articles, lectures and book chapters.
Much of his work investigated differences in intelligence
among social classes, gender and race, and
examined heritability among intelligence of identical
twins reared apart. It was this work that was to create
the controversy.
During the 1970s Leon Kamin, who was opposed to
the idea of genetic heritability, and Arthur Jensen, who
was in favour of genetic heritability, both spotted something
in Burt’s reports on correlations for IQ test scores
of identical twins (Jensen, 1973; Kamin, 1974). Burt’s
original results were published in 1943 for 15 pairs of
twins. In 1955 he had results for 21 pairs of twins (including
the 15 original pairs of twins), and by 1966 he
reported the results for a total of 53 pairs of twins. What
Kamin and Jensen noted was that the correlation coefficients
between the IQ scores were similar across the
samples. In his studies, Burt reported the following coefficients:
r ϭ 0.770 (1943); r ϭ 0.771 (1955); and r ϭ
0.771 (1966). Kamin and Jensen suggested that they
would expect to see greater variability among the correlations
when more sets of twins were added, and that it
was very unusual that the correlations had stayed the
same.
On 24 October 1976, an article, ‘Crucial data was
faked by eminent psychologist’, was published by
Dr Oliver Gillies in the UK newspaper The Sunday Times.
In this article, Gillies not only pointed to the problems
with the correlational data but also claimed that Burt had
invented his co-authors, Miss Margaret Howard and
Miss J. Conway (who Burt claimed helped him update
the twin data), as they could not be traced. Gillies also
pointed out that Howard and Conway did appear to have
written book reviews praising Burt’s work for the Journal
of Statistical Psychology – but only during the time that
Burt was editor. Later, US psychologist William H. Tucker
picked up on accusations that Burt would have been unlikely
to find so many sets of monozygotic twins reared
apart. Tucker (1997) examined the numbers of participants
involved in twin studies with similar criteria used
by Burt between 1922 and 1990 and found that the combination
of all the twins together did not total the number
of participants Burt had obtained.
With this, Cyril Burt seemed to be forever denounced
as a fraud. However, people have reviewed this evidence.
For example, UK psychologist Ronald Fletcher,
(1991) set out to locate the missing research assistants.
He found proof of their existence. One of the assistants,
Miss J. Conway, worked in child care for London County
Council. Miss M. A. Howard is listed as a member of
the British Psychological Society in 1924. In addition,
Fletcher reported that other individuals stepped forward
and said they remembered Burt’s assistants. Fletcher
suggests that this repeated accusation against Burt
may be the result of poor investigatory journalism rather
than imagined helpers.
This is not the only accusation against the media.
Some authors such as Fletcher, Robert Joynson (1989),
and J. Phillipe Rushton (1994) have suggested that
Burt’s reported correlation coefficients indicated a
strong relationship between genetics and intelligence –
and they point out this was not, and still isn’t, a popular
finding. This is because such findings have strong implications
where differences in intelligence are sexually or
racially based. Often, there is the inference that certain
groups of people (based on sex or race differences) are
genetically inferior in intelligence. We will explore such
ideas in full later in the chapter. However, these ideas
are surrounded by controversy, and Fletcher and
Rushton point out that, if Burt’s findings can be discredited,
then it seems to some that the case for the genetic
heritability of intelligence is discredited. Therefore,
Fletcher and Rushton argue that people who disagree
with the genetic heritability of intelligence will be more
likely to conspire in the media and in academia to discredit
Burt’s work.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 345
346 PART 2 INTELLIGENCE
Biological variables
and maternal effects
School and education
Culture
Decontextualisation
Quantification
Biologisation
Nutrition
Lead
Prenatal factors
Family environment
Shared and non-shared
environments within- and
outside-family factors
Social class and
socioeconomic status
Birth order and
family size
Figure 14.3 Environmental effects on intelligence.
Rushton points to possible conspiratorial behaviour.
He suggests that not only the media sought to discredit
Burt’s findings too quickly, but other academics did,
too. He cites the example of Burt’s papers being destroyed
by his housekeeper immediately after his death –
on the advice of Professor Liam Hudson, an educational
psychologist at Edinburgh University, an ardent opponent
of Burt who had rushed to Burt’s flat after learning
he had died.
Rushton also points out that new evidence from studies
of twins raised apart (as we see from the summary of
Ridley earlier in this chapter) indicates results similar to
Burt’s high heritability estimate. Therefore, the accusations
that he simply made up all the data to exaggerate
the effects of genetics are somewhat weakened.
Authors like Ronald Fletcher and Robert Joynson
suggest that the case of fraud by Burt is not proven.
Whatever the truth, and the lessons to be learnt from
Cyril Burt, we hope that whatever judgements you
make, you understand the importance of giving the full
facts before you pass on the story . . . and never make
up your data.
weight at birth, height, years in school, father’s economic
status, father’s and mother’s education and influence, average
TV viewing, self-confidence, criminality and emotional
adaptation. However, the American Psychological Association
task force (Neisser et al., 1996) identified four main
areas of environmental effects on intelligence which can be
deemed as the most important. This information allows us
to concentrate the debate. Some of these environmental
effects you will have come across before (for example, nutrition,
schooling and occupation) in the last chapter, but
we will revisit them so that you can see how they apply directly
to the issues of the nature versus nurture debate on
intelligence. These four areas are (1) biological variables,
(2) family, (3) school and education and (4) culture (see
Figure 14.3).
Biological variables and
maternal effects
It is well acknowledged by the APA task force (Neisser et al.,
1996) that a number of biological factors influence intelligence.
These include nutrition, before- and after-birth factors
and substances like alcohol and lead poisoning.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 346
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 347
Nutrition
We introduced much of the nutrition hypothesis in the last
chapter. Nutrition is the study of food; specifically, the relationship
between diet and states of health and disease. You
will remember from the last chapter that Lynn (1990) has
proposed that nutrition and healthcare improvements are
among the main factors of the Flynn effect. However, outside
Lynn’s hypothesis there are findings that nutritional
sources can aid aspects of intelligence.
Australian nutritionist Wendy H. Oddy and her colleagues
(Oddy et al., 2004) examined over 2,000 Australian
children and followed them from birth until the age of 8
years. Oddy and her colleagues found that stopping breast
feeding early (at 6 months or less) was associated with reduced
verbal intelligence, while children who were fully
breast-fed for more than 6 months scored between 3 and 6
IQ points higher on a vocabulary intelligence test than did
those children who were never breast-fed. Similarly, another
study conducted by US health psychologist Melanie
Smith and her colleagues (Smith, Durkin, Hinton,
Bellinger and Kuhn, 2003) examined 439 school-age, lowbirth-weight
children born in the United States. These authors
found differences in IQ test scores between breast-fed
children and those who did not receive any breast milk.
These were 3.6 IQ points for overall intellectual functioning
and 2.3 IQ points for verbal ability. You will also remember
the debate in Chapter 13 surrounding the study
carried out by two Welsh psychologists, David Benton and
Gwilym Roberts (Benton and Roberts, 1988), who found
that children given a vitamin-mineral supplement containing
several vitamins and minerals were found to show increased
IQ scores.
Regardless of the controversy surrounding this work,
findings do suggest that nutrition has a positive effect on
intelligence. It is important to remember, though, that a
number of socioeconomic conditions are often associated
with nutrition. You may also remember the words of Lynn
(1990), who we discussed in the last chapter: the nutrition
hypothesis sees nutrition as a package (or nurturing environment)
in which increased intelligence is part of a nurturing
environment that includes increased height and life
span, improved health, decreased rate of infant disease and
better vitamin and mineral nutrition. Where those things
do not occur – where there is poverty, malnutrition and
low economic and social opportunities – there might be
lower intelligence scores.
Lead
However, while nutrition is seen as having a positive effect,
there are occasions when other biological factors can have a
negative effect on IQ. Neisser et al. (1996) highlight
research concentrating on the effect that exposure to lead
can have on intelligence.
The most comprehensive study was carried out in a
place called Port Pirie in Australia. In 1986, Australian
psychologist Anthony J. McMichael and four colleagues
(McMichael, Vimpani, Robertson, Baghurst and Clark,
1986) examined the possible relationship between body
lead burden and pregnancy outcome among 749 pregnant
women in, and around, the largest lead smelting facilities in
Australia. Among these women, premature deliveries were
statistically significantly associated with higher levels of
maternal blood lead concentration at delivery. What followed
the initial findings was a series of studies in Port
Pirie, which looked at the association between environmental
exposure to lead and children’s intelligence at 2 years
(McMichael et al., 1988), 4 years and 7 years (Baghurst et al.,
1992) and 11 and 13 years (Tong, Baghurst, McMichael,
Sawyer and Mudge, 1996). The Port Pirie cohort study
started in 1979 and involved 723 children, though 375 took
part in the final study. IQ scores were made on the Bailey
IQ scales at 2 years of age, the McCarthy IQ scale at 4 years
of age and the Wechsler IQ scale for children at 7 years,
11 years and 13 years of age. At all these ages, IQ scores were
significantly associated with lead concentration in people’s
bodies, even when socioeconomic status, home environment
and maternal intelligence were controlled for. This
study suggests that there is an association between early
exposure to environmental lead and intelligence, and it
persists into later childhood.
Prenatal factors
Finally, there are prenatal factors. You are well aware that
pregnant women are expected to stop drinking and smoking.
According to many health councils, avoiding smoking
and alcohol consumption in pregnancy is crucial. Smoking
nearly doubles a woman’s risk of having a premature or
low-birth-weight baby who faces an increased risk of serious
health problems. Further, many conditions can arise
from the mother’s alcohol consumption when she is pregnant;
the most common condition is foetal alcohol syndrome
(FAS), which is characterised by a pattern of facial
abnormalities, growth retardation and brain damage.
Neisser et al. (1996) point to these types of consumption
by mothers during pregnancy as having an effect on
intelligence. Low-birth-weight babies, as well as babies suffering
from FAS, show reduced intelligence. Danish scientist
Erik Lykke Mortensen and colleagues at the University
of Copenhagen (Mortensen, Michaelsen, Sanders and
Reinisch, 2005) examined maternal smoking and subsequent
IQ scores among 3,044 males aged between 18 and
19 years. The study found that regardless of factors such as
parental social status and education, single-mother status,
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 347
mother’s height and age, number of pregnancies, the
women who smoked 20 or more cigarettes daily late in
their pregnancy were likely to have sons who performed
less well on standardised IQ tests at age 18 or 19. Evidence
also suggests that FAS is related to a number of cognitive
functions. US psychologist Sarah Mattson and colleagues
at San Diego State University (Mattson and Riley, 1998;
Mattson, Riley, Delis, Stern and Jones, 1996) have found
that children prenatally exposed to alcohol exhibit a variety
of problems with memory (when they found that children
with FAS aged 5 years to 16 years had learned fewer words
than children of comparable ages) and demonstrate attention
problems. Uecker and Nadel (1996) found that children
of mothers who drank heavily during pregnancy
performed badly in learning spatial relationships among
objects. Furthermore, South African psychologist Piyadasa
Kodituwakku and her colleagues (Kodituwakku, Handmaker,
Cutler, Weathersby and Handmaker, 1995) have
shown that children with FAS show deficits in activities
that require abstract thinking, such as planning and organising
information.
However, research is by no means conclusive. Other research
suggests that the links between factors such as
smoking, or alcohol consumption, in pregnancy and intelligence
might be moderated, disappear or be highlighted by
other factors. US psychologists S. W. Jacobson, J. L. Jacobson,
R. J. Sokol, L. M. Chiodo and R. Corobana (2004)
found, among 337 inner-city African American children,
that prenatal alcohol exposure was not related to IQ scores
on the Weschler Intelligence Test. However, they found that
among children who had older mothers, prenatal alcohol
was related to IQ scores on the Wechsler Intelligence Test.
Bailey et al. (2004) examined alcohol use among mothers
at a prenatal visit, and then IQ among children at 7 years,
among 500 black children. Again, no relationship was
found between prenatal alcohol exposure and intelligence,
though mothers who binge drink when pregnant were
1.7 times more likely to have children who had IQ scores in
the mentally retarded range.
These findings suggest that age and excess drinking are
further factors in the relationship between prenatal drinking
and offspring intelligence. But overall, the findings suggest
that smoking and alcohol consumption are factors
that, to a greater or lesser extent, are connected with IQ.
Maternal effects model
Today, factors such as prenatal nutrition and alcohol consumption
are combined into the maternal effects model.
US psychiatrist Dr Bernie Devlin and US statistician
Michael Daniels (Devlin, Daniels and Roeder, 1997)
showed that prenatal conditions may have substantial
effects on the concordance of subsequent scores on IQ for
identical twins. Previously, maternal effects had usually
been assumed to be small, or non-existent, in terms of
affecting genetic variance of intelligence between twins.
However, a meta-analysis of 212 studies suggests 20 per
cent of genetic variance between twins and 5 per cent between
siblings. These authors suggest that broad heritability
estimates when including maternal effect (additive and
non-additive genetic variance) might have to be reduced
from about 60 to 48 per cent. This suggestion indicates that
the environmental effects on intelligence may extend to
interactions with biological factors.
Family environment
The second environmental factor identified by Neisser
et al. (1996) is the family environment. There are three
sources of related research and evidence that we will concentrate
on in this discussion:
shared and non-shared environments – within- and
outside-family factors;
the social and economic status of the family and the
intelligence of the child;
birth order, family size and child intelligence.
Shared and non-shared environments
We saw in the last discussion that the conception of genetics
as a single dimension has developed over time. The
same could be said of environmental factors. Within behavioural
genetics, the conception of how the environment
influences intelligence is through two sets of experiences:
shared and non-shared. When growing up, siblings (brothers
and sisters) are thought to experience both shared and
unique environments. Shared environments are environments
that are shared between two individuals, while
non-shared environments are environments that are not
shared between two individuals who share genes. Siblings
growing up within the same family will share many environments.
These environments may range from minor experiences
to more significant ones. Therefore, two siblings
having the same parents, living within the same house,
going to the same school and experiencing particular times
together (e.g., same family relatives, home environment,
chaotic mornings before school, dad’s awful jokes) have
shared environments. A unique environment is an environment
that has not been shared by siblings. Again, these
environments may range from minor experiences to significant
ones. Examples of unique environments might be
when two siblings have been raised by different families.
However, siblings raised in the same family might also have
unique environments from each other. Siblings may have
348 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 348
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 349
Child is
Intelligent
Child is
Intelligent
Child is
Intelligent
Intelligent
behaviour
Child then learns
more intelligent
behaviour
Parent responds to the
behaviour by teaching
child something new
Child being intelligent
causes the parent to
be intelligent
Child shows
intelligent behaviour
Child shows some
intelligence
Child is intelligent
no matter the
reaction of the
parent
Child is intelligent
as the result of
shared genes
Child’s intelligence
in response to
parent’s reaction to
their intelligent
behaviour
Passive
model
Child-effects
model
Parent-effects
model
ChildParent
different sets of friends, may go to different schools, may
have different types of relationships with their parents and
have different interactions with teachers.
What is important in this area is that the theory and
research around the differences between environment influences
on intelligence has grown in complexity. To begin
with, researchers tend to concentrate on comparing how
shared and non-shared environmental factors influence
on intelligence. Early consideration by reviewers such as
Bouchard (1994) and Eysenck (1990b) suggested that the
environmental influences shared by siblings or twins contribute
only marginally to intelligence differences. However,
one interesting point to emerge from the literature,
carried out by such researchers as US behaviour geneticists
Braungart, Plomin, DeFries and Fulker (1992a), is that
those environmental factors that are unique (non-shared)
to family members are influential over shared environmental
factors. Therefore, non-shared environmental factors,
such as different peer friendships, are important mechanisms
that explain why members of the same family may
differ in their intelligence. This idea is supported by two
pieces of research suggesting that the extent of differences in
the experiences during childhood among siblings have been
found to be related to intelligence differences in adulthood
(Baker and Daniels, 1990; Plomin and Daniels, 1987).
Such a finding has led to the development of whole
areas of research that have emphasised how important
non-shared environmental factors are to intelligence. Most
of the research in this area considers how non-shared environmental
factors develop (1) within the family and (2)
outside the family.
Within-family factors
US behavioural geneticist David Reiss (1997) identifies three
ways in which inherited genes form phenotypes (behaviours)
based on the family environment (see Figure 14.4):
the passive model
the child-effects model
the parent-effects model.
First is the passive model. This model suggests that intelligence
is generally explained by the 50 per cent overlap
between a child and their parent. Therefore, intelligence
may occur in the child because the child and parent share
Figure 14.4 Reiss’ three models of genetic transmission.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 349
the same genes that influence a particular type of behaviour.
For example, if a child is highly intelligent due to genetic
influences, they are so because one of their biological
parents had the genes that cause this high intelligence. The
model, very much, just assumes a general genetic overlap
and inheritance of behaviour without considering other
possible factors and interactions within the family. This is
why it is called the passive model. The other two models
very much emphasise other dynamic occurrences.
In the child-effects model, the genes cause intelligence
in the child, which in turn causes the same or similar
behaviour in the parent. Within this model, the parent does
not matter in the development of the behaviour, as the
child’s development of intelligence is the result of genes.An
example of this is that the shared genes cause the child to
be intelligent (due to their genetic makeup), which in turn
causes the parent to act intelligently back to the child (due
to their genetic makeup). The parent’s own intelligence
does not matter in the development of the behaviour, as
the child’s intelligence is a consequence of genetic makeup
of the child rather than the parent.
US psychologist Judith Harris (1995) has expanded this
viewpoint to child-driven effects, which influence family
circumstances that in turn influence the child’s intelligence.
Harris documents studies showing that adults do
not behave in the same way to a child who shows different
tendencies. They will treat an attractive child differently to
one of their children who is less attractive; they will react
differently to the child who shows bad behaviour than to
the one who is well behaved. This behaviour will also
apply to intelligence. For example, imagine a family with
twin children, one of them intelligent and the other not as
intelligent. These differences in the children will cause different
reactions in the parents. The parents will begin to
treat their children differently. The intelligent one may be
encouraged to engage in more intelligent activities, while
the one perceived to be less intelligent might not be encouraged
to do these activities. Harris suggests these reactions
by parents to their children’s natural intelligence
tendencies can be viewed in two ways: positive feedback
loops and negative feedback loops. Positive feedback loops
arise from parents reinforcing children’s natural tendencies,
as in the example described earlier, where children’s
natural intelligence abilities are encouraged and any differences
between children in their intelligence are developed
(the intelligent child is encouraged to be intelligent,
while the ‘unintelligent’ child is encouraged and allowed to
engage in other activities). Negative feedback loops occur
when children are stopped from behaving in ways consistent
with their natural tendencies. Therefore, an intelligent
child might be encouraged to stop engaging in merely
intelligence-stimulating activities, and an ‘unintelligent’
child might be encouraged to spend more time engaging
in such activities.
In the parent-effects model, the behaviour of the
child is responded to by the parent, which in turn brings
out behaviour in the child (see Figure 14.4). Within this
model, how the parent responds does affect the development
of the child’s behaviour. For example, a child may
begin showing intelligence; this then leads the parent to
show intelligence with the child (as it is part of their genetic
makeup). This in turn causes the child to become even
more intelligent (as it is part of their genetic makeup).
Within this model, the parent’s behaviour leads to the development
of intelligence, which then leads to the development
of intelligent behaviour in the child.
Again, Harris extends this idea to within-family situations.
In these situations, children might be treated in a
particular way by parents, not because of that child’s own
characteristics, but because of the parents’ own beliefs or
the characteristics of a child’s siblings (brothers or sisters).
Let us first look at the example of how the parents’ own
beliefs shape natural tendencies of children. Again, take
our family with the one intelligent twin and the one ‘lessintelligent’
twin. Our parents of the family may have certain
beliefs about behaviour, such as ‘children should
know the limits of their knowledge and never contradict
their parents’. Therefore, the children will be encouraged
and directed to behave in such ways. So, in our case of the
intelligent and less-intelligent twin, the intelligent child
who is likely to contradict their parents will be encouraged
not to show their intelligence, and the unintelligent child
will be encouraged to gain some knowledge up to their
‘limits’. Both children will have had their new behaviour
(knowing their limits and not contradicting their parents)
driven by their parents’ behaviour. Secondly, let us look at
how parents might influence children’s behaviours in
terms of a child’s siblings. Harris notes research suggesting
that parents who consider their first child to be ‘difficult’
tend to label their second-born as ‘easy’. We can also see
how less-intelligent children might be asked, or encouraged,
to be more like their intelligent sibling. Equally, the
intelligent sibling might be encouraged to spend less time
in intelligence-promoting activities and play more like
their brother or sister.
What Reiss’ and Harris’ commentaries do is to suggest
that within-family effects pose problems when considering
genetic heritability. That is, child effects and parent effects
can lead to overestimations and underestimations of heritability.
Remember, behavioural geneticists looking at intelligence
are only looking at the concordance between sets
of children based on their scores on an intelligence test at
some point. However, let us return to our family with one
intelligent twin and one less-intelligent twin. Let us imagine
that the parents of these twin children have been engaged
in a negative feedback loop. They have been trying
to encourage both children to be similar, that is, somewhat
intelligent. Therefore, the intelligent child has been dis-
350 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 350
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 351
couraged from being intelligent all the time, and the lessintelligent
child is being discouraged from being too unintelligent.
If we then compared these two children, we would
find that these twin children have similar intelligence; but
this is, in fact, not due to genetic tendencies at all. It is simply
because the parents are trying to encourage similar behaviour
in both children (i.e., not too active or not too
quiet). Therefore, any estimation of similarities in intelligence
being due to genetic heritability of the twins would
be an overestimation. However, if the same pair of twins
had been reared differently and both had been in a positive
feedback loop – the intelligent child had been encouraged
to be more and more intelligent and the less-intelligent
child had been encouraged to be more and more unintelligent
– then any estimation of the similarities in intelligence
being due to genetic heritability would be an underestimation
because the differences have been exaggerated due to
parents encouraging the twins to be more and more like
themselves. Therefore, as Harris concludes, children’s
within-family situations not only play an important role in
shaping their intelligence but also are an important consideration
in estimating the genetic heritability of intelligence.
Outside-family factors
However, Harris (1995) has suggested that non-shared factors
outside the family may in fact be more important in
developing people’s intelligence. Harris presents the group
socialisation theory to explain the importance of nonshared
environmental factors in determining intelligence.
Group socialisation theory is based largely on the ideas
surrounding social identity theory and social categorisation
(Tajfel and Turner, 1986). Social psychologists have
provided a lot of theoretical and empirical research work
that has looked at how individuals perceive their social
world as comprising in-groups and out-groups and suggesting
that these groups help us form our social identity.
Social psychologists argue that one mechanism humans
use for understanding the complex social world is social
categorisation. In social categorisation, individuals are
thought to place other individuals into social groups on the
basis of their similarities and differences to the individual.
Put simply, individuals who are viewed as similar to the
person tend to be placed within their in-group. Individuals
who are viewed as different to the person tend to be placed
within an out-group. This is a process by which we come to
understand our world. As a consequence, the individual’s
identity (social identity) is based on and derived from the
groups they feel they belong to and their understanding of
their similarities and differences to different social groups.
Social groups can be anything, but common groups could
be sex group, ethnic group, your religion, your peers, your
interests, your educational status and so on. As such, your
identity is based, to greater or lesser extent, on how much
you identify with different social groups. What is also important
to our identity is that when we attach ourselves to
certain groups, we also try to fit in with those groups;
therefore, our intelligence might begin to reflect the characteristics
of a certain group (i.e., you might make friends
with people who are highly intelligent; you may then do
more intellectual activities than you used to, and you may
become more intelligent).
Harris uses this theoretical basis to show how social
groups can influence people’s intelligence and points out
how these non-shared environments that occur in children
of the same family can have a huge effect on intelligence. As
part of this theory, Harris lists five aspects that are important
to consider in how non-shared characteristics might
influence our intelligence (we go into greater depth on each
of these issues in Chapter 8 if you want to read more).
Context-specific socialisation – This aspect refers to
the fact that children learn intelligence abilities not only
at home but also outside the home. As children get older,
they become less influenced by their family life and more
influenced by their life outside the family home.
Outside-the-home socialisation – In this aspect Harris
makes the point that children may identify with a number
of social groups, based on people’s age, gender, ethnicity,
abilities, interests, personality, intelligence etc. In
other words, we have a range of groups that we identify
with and share norms with (attitudes, interests, intelligence),
and these groups have different influences on
our intelligence.
Transmission of culture via group processes – In this
aspect Harris makes two points about the transmission
of culture via group processes, which establish norms in
our social world that influence our intelligence. First,
the shared norms that might influence a child’s intelligence
aren’t necessarily the result of parents sharing
them with their children; they are really the result of
shared norms among the parents’ peers and social
groups being passed on to the children. That is, your
parents’ values, abilities, personality and intelligence are
not the result of their parents’ norms, but rather of their
own social identity, their identification with their own
social groups. The second point considers that individual
norms, which we have developed from our family,
are shared with other people only if they are accepted.
For example, an individual might be intelligent and like
reading books. However, unless their friends approve of
this behaviour, they may be unlikely to carry on this
pastime.
Group processes that widen differences between social
groups – It is important to note that with your intelligence,
norms are based not just on how you identify
with your in-group but also on whether you identify
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 351
with or reject the out-groups. For example, consider sex
roles; your intelligence as a male or female isn’t just
based on your identification with people of your own
sex, but on your rejection of characteristics of the opposite
sex. For example, some young women at school believe
that subjects such as science and mathematics are
men’s subjects; consequently, these women become interested
in what they perceive to be women’s subjects,
such as English. This behaviour might have an effect on
intelligence.
Group processes that widen differences among individuals
within the group – So far, we have assumed
that all the groups that we are involved in basically share
the same structure. However, we know that within all
our social circles we play different roles that might influence,
or bring out, different aspects of our intelligence.
In our family, as a child, we take a less-senior role; with
our friends we might be allowed to be more like ourselves.
However, the opposite may be true, and we might
not feel we lead a group of friends, but tend to do what
others say. It may even be possible that among one
group of friends you feel more comfortable than others.
Harris’ point is that our position in groups changes,
and that our intelligence – and influences on our intelligence
– change due to the hierarchies within a group.
For example, if you are in a group of friends and they all
look up to you, your intelligence will be influenced
because you might think there is an expectation to come
up with ideas for things to do or to solve problems
within the group.
What is important to consider in both within-family
and outside-family factors is that these aspects can influence
intelligence of children to a much greater extent than
previously thought. It is not Harris’ point that behavioural
genetics is wrong and that environmental factors are more
important, but rather that behavioural geneticists may
sometimes have oversimplified family influences. By ignoring
these variables, behavioural geneticists might be underestimating
or overestimating heritability effects of either
genetics or the environment on intelligence.
Socioeconomic status of the family
A family’s socioeconomic status is based on its income,
parental education level, parental occupation and status in
the community. Socioeconomic status is related to various
factors, including number of children in the family, opportunities
for success in employment, health and area of residence.
These factors all might influence intelligence.
Often socioeconomic status is ranked in many countries.
For example, in the United Kingdom, one way in
which socioeconomic status is measured is by the grading
of parents’ occupations into five categories:
Class I: Professional occupations;
Class II: Managerial and technical occupations;
Class III: Skilled occupations: Manual (M) and unmanual
(U);
Class IV: Partly skilled occupations;
Class V: Unskilled occupations.
Socioeconomic status is related to intelligence. Authors
such as Linda Gottfredson (Gottfredson, 1986), Arthur
Jensen (Jensen, 1993a), Richard Herrnstein and Charles
Murray (Herrnstein and Murray, 1994) and J. Phillipe
Rushton and C. D. Ankney (Rushton and Ankney, 1996)
estimate that in Europe, North America and Japan, socioeconomic
status is significantly correlated with scores
on standard IQ between r ϭ 0.3 and r ϭ 0.4, and that there
are 45 IQ points between members of the professional
occupations (Class I) and those of unskilled occupations
(Class V).
In addition, Nicolas Mackintosh (1998) and Nicholas
Mascie-Taylor (1984) presented evidence linking socioeconomic
status to intelligence using the British National
Child Development Study (NCDS) data. The NCDS examined
social and obstetric (care of women during and after
pregnancy) factors associated with stillbirth and infant
mortality among over 17,000 babies born in Britain in
1958. Surviving members of this birth cohort have been
surveyed on five further occasions in order to monitor
their changing health, education, social and economic circumstances,
including IQ, in 1965 (age 7), 1969 (age 11),
1974 (age 16), 1981 (age 23) and 1991 (age 33). Mackintosh
(1998) and Mascie-Taylor (1984) present evidence that
even when aspects such as financial hardship, birth weight,
size of family, overcrowding, type of accommodation and
residence area are taken into account, children who had fathers
in Class I (professional occupations) scored 10 IQ
points higher than did children who had fathers in Class V
(unskilled occupations).
Research also suggests that improved socioeconomic
status can improve intelligence. Canadian behavioural geneticist
Douglas Wahlsten (Wahlsten, 1997) points to a
series of adoption studies in France, in which an infant is
moved from a family having low socioeconomic status to a
home where parents have high socioeconomic status, and
the child’s IQ score improves by 12 to 16 points. Wahlsten
also points to studies in the United States that have demonstrated
improvements in children’s IQ by the same margin,
achieved by improving the lives of infants in families with
low educational and financial resources and providing
them with additional educational day care outside the
home, every weekday from the age of three months to the
start of school.
352 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 352
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 353
These types of findings turn our attention to the research
that has examined the conditions arising in certain
families and how those conditions have influenced intelligence.
One interesting and extensive debate has arisen from
studies examining the influence on intelligence of birth
order and family size.
Birth order, family size and intelligence
In 1973 Dutch psychologists Lillian Belmont and Francis
Marolla (Belmont and Marolla, 1973) published a study
looking at the birth order of the child, the size of family to
which the child belonged and the child’s overall IQ score
on the Raven Progressive Matrices, among 386 nineteenyear-old
Dutch men. What Belmont and Marolla (1973)
found was that even when social class was controlled for,
children from larger families had a lower IQ. Furthermore,
the authors found that within each family size, the firstborn
child always had a better IQ; and to some extent there
were declining scores with rising birth order, so that the
first-born children scored better than second-born children,
second-born children scored better than third-born
children and so on. These authors also found that these two
factors interacted, and as family size increased and birth
order position increased, IQ scores became lower. So for
example, a second-born child from a family with three
children would score higher than would a second-born
child from a family containing four children.
Clearly, birth order, family size and intelligence are of
interest to parents, politicians and researchers. Furthermore,
if such findings are correct, they would have implications
for optimum family size and parental choice
regarding children’s education. Since then, hundreds of research
articles have addressed the relationship between
family size, birth order and intelligence, and the proposed
relationships between family size, birth order and IQ have
been found among many cross-section studies. For example,
Russian psychologist T. A. Dumitrashku (Dumitrashku,
1996) found that family size and birth order affected intelligence
among Russian schoolchildren.
However, this wouldn’t be a section about intelligence if
debate didn’t fiercely surround these findings. The debate,
today, on family size, birth order and intelligence centres
on the explanation of why, and whether, such effects occur.
Family size and intelligence
In the research area of family size and intelligence, Joseph
Lee Rodgers and his colleagues (Rodgers, Cleveland, van
den Oord and Rowe, 2000) published a seminal paper that
looked at data from the United States. The National Longitudinal
Survey of Youth (NLSY) followed 11,406 young
people at yearly intervals from ages 14 to 22 years, and then
children born to the original female respondents were
surveyed every other year. Rodgers et al. found no direct
relationship between family size and intelligence. They suggested
that previous research has been inaccurate because
Often evidence of high levels of ability can be found in unexpected places.
Source: Getty Images/Daniel Berehulak
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 353
it combined ‘across-family’ measures (family size) with
‘within-family’ (birth order) measures and treated them in
the same way. That, is previous authors had treated family
size as a within-family effect.
Let us explain what the authors mean. What the authors
are highlighting, here, is a statistical fallacy (see Chapter 26
on academic argument) that occurs when comparing populations
of people. Say, for example, that a statistical agency
released figures for death rates of the UK Army during the
recent Iraq War and for death rates in London (the capital
of the United Kingdom). The agency found that among the
army, death rates were 13 per thousand, while deaths in
London were 26 per thousand. You perhaps also would not
be surprised to find that the announcement of these figures
by the statistical agency caught the attention of the media.
You might even find a national newspaper running a headline
story suggesting that people were safer in the army in
Iraq than they were living in London. However, the problem
with this sort of statement is that you are comparing
two different populations. In the army population you have
men and women who are healthy, and most of them are
young. In the second population, London, you have a full
age range of people – including those people with high
mortality rates, such as old people, and people who are terminally
ill. The issue is that you are comparing two populations
for which a number of different factors determine
death rates.
This fallacy applies to the current debate. Rodgers et al.
illustrate the fallacy with this example. They suggest comparing
the intelligence of three children, but these children
are:
a first-born child in a large middle-class white family in
Michigan;
a second-born child in a medium-sized affluent black
family in Atlanta;
a third-born child in a small low-income Hispanic family
in California.
If differences are observed in these children’s intelligence,
Rodgers et al. suggest it is impossible to tell whether
the differences are down to birth order, family size, socioeconomic
status, region of the country or any other variables
related to these dimensions.
Birth order and intelligence
In the area of birth order and intelligence, research still
generally supports Belmont and Marolla’s findings. However,
other authors have sought to explain that this relationship
may be an artefact of another relationship rather
than a real relationship. There are three models explaining
why birth order may be linked to intelligence: the admixture
hypothesis, the confluence model and the resource
dilution model.
The first is the admixture hypothesis. E. P. Page and
G. Grandon (1979) and more recently Joseph Rodgers
(2001) suggested an ‘admixture hypothesis’ that explains
the relationship between birth order and IQ. What this hypothesis
suggests is that parental intelligence, or socioeconomic
status, are additional factors to consider in the
relationship between birth order and IQ scores, coupled
with the fact that parents with lower IQ scores tend to have
more children. This has made findings in previous studies
look as if higher birth order causes lower intelligence, when
in fact lower intelligence results because parents with lower
socioeconomic status and IQ scores tend to have more children.
For example, a parent with five children is likely to
have a lower IQ score and a lower socioeconomic status.
Parents with higher IQ scores and higher socioeconomic
status tend to have fewer children. Consequently, any calculation
of the relationship between birth order and intelligence
is problematic because parents with higher IQ scores
and higher socioeconomic status do not tend to have as
many children. Thus there cannot be equal measurement
of the number of children across the population.
The second model is the resource dilution model.
This model was proposed by Judith Blake (Blake, 1981) and
elaborated by Douglas Downey (Downey, 2001), but its
ideas were first presented by Galton (1874). The resource
dilution model of birth order and intelligence test scores
suggests that parental resources (time, energy and financial
resources) are finite (i.e. not endless) and that, as the number
of children in the family increases, the resources (time,
energy and financial) that can be gained by any single child
decreases. Therefore, the first child will get 100 per cent of
available resources from their parents, the second child will
only ever get 50 per cent, the third child will only ever get
33 per cent and a fourth child will only ever get 25 per cent.
This model also feeds into the idea that children in larger
families have lower intelligence test scores because, as that
family grows, the resources that can be accessed also di-
minish.
Third is the confluence model, which was originally
proposed by US psychologist Robert B. Zajonc (Zajonc,
1976; Zajonc and Markus, 1975) – though, again, Galton
(1874) proposed some of these ideas. The confluence
model suggests that intellectual development, and therefore
intelligence, must be understood within the context of
the family, and there is an ever-changing intellectual environment
within the family. Zajonc suggests the following
factors might influence the relationship between birth
order and intelligence:
First-borns have the advantage of some time in which
they do not have to share their parents’ attention with
any of their siblings.
Any additional birth automatically limits the amount of
attention any of the siblings get, including the first-born.
354 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 354
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 355
First-borns and older siblings have to look after and care
for younger siblings to some degree. This means that
they undertake some amount of responsibility and may
have to explain things to their younger siblings. Zajonc
believed that this sort of tutoring helps the older children
to develop intelligence abilities, as they have to explain
ideas and processes to other people.
First-born children are exposed to a greater proportion
of adult language and ideas from their parents. Those
children born later are exposed to less-mature speech
and ideas because they listen not only to their parents,
but to their other siblings. This means they spend a
lower proportion of their time listening to adult language
and ideas and a greater proportion of time listening
to other children’s language and ideas.
These last two findings also feed into the idea that children
in larger families have lower IQ scores because, as that
family grows, the agenda and the context of the family
focuses more and more on the children.
Education and intelligence
We mentioned education and its relationship to intelligence
in the last chapter. However, we will remind you of
some of the findings and extend your view on this area in
this chapter.
Neisser et al. (1996) found that education is both an independent
and dependent variable in terms of its relationship
to intelligence. Going to school is likely to increase
your abilities, particularly those that comprise intelligence
(intelligence is a dependent variable), and intelligence is
likely to influence your attendance at school and your
length of schooling (i.e., whether you end up going to university)
and the quality of school you attend (intelligence is
an independent variable here). Consequently, intelligence
and education are intrinsically linked.
You will remember evidence from the last chapter.
Overall, reviews by US psychologists Alan Kaufman and
Elizabeth Lichtenberger (Kaufman, 1990; Kaufman and
Lichtenberger, 2005) provide a review of key papers that
have looked at the correlation between general intelligence
and school attainment and achievement. They conclude
that the average correlation between IQ scores and a number
of school indicators is around r ϭ 0.50, suggesting that
intelligence does predict performance at school. Also, two
academics of the Hebrew University in Jerusalem, Sorel
Cahan and Nora Cohen (1989), compared the effects of a
year of school (controlling for age) with those of a year of
age on a number of verbal (e.g., verbal and numerical
skills) and non-verbal (abstract and reasoning tests, including
the Raven’s Matrices) intelligence tests. Length of
schooling was important in predicting performance, and
mattered more than age, for all the verbal tests. Length of
schooling, however, made a contribution – but a smaller
contribution – to performance on some of the nonverbal
tests, including the items from the Raven’s Matrices.
Other key evidence you need to know when considering
education as an environmental factor on intelligence is
found in the well-cited papers of US child development
psychologist Stephen Ceci (1990, 1991). Ceci did a metaanalysis
of studies, and his findings suggest that there are
many effects of education on intelligence test scores. The
data presented by Ceci includes the overall finding that
children who attend school regularly score higher on intelligence
tests than do those who attend less regularly, intelligence
test scores among pupils decrease over the long
summer holidays and there is a rise of 2.7 IQ points for
each year of schooling (see also Winship and Korenman,
1997). Douglas Wahlsten (1997) notes that studies have
shown that delays in starting school cause intelligence tests
scores to drop by 5 IQ points a year (i.e., Winship and
Korenman, 1997).
Also, it is worth reminding you of Head Start and other
similar programmes that have explored the relationship
between education and intelligence. You remember that
Head Start, which we mentioned in the last chapter, was
started in the 1960s by President Lyndon Johnson and was
designed to give America’s poorest children a head start in
preparing them for school and to start to break the cycle of
poverty. Evidence was provided to assess the usefulness of
such a programme. US individual differences psychologist
Charles Locurta (Locurta, 1991) provides a review of this
evidence. In 1969, Arthur Jensen suggested that Head Start
had failed. The reason for Jensen’s pronouncement was
that, although children attending the programme showed
an initial increase in IQ points – sometimes as much 7–8
points on IQ tests – after 2 or 3 years these higher IQ points
were lost (Locurta, 1991). There has been a lot of debate
about the effectiveness of Head Start, but in terms of IQ
gains, McKey et al. (1985) reported that children enrolled
in Head Start had significant immediate gains in IQ scores.
However, in the longer term (3–4 years), the IQ test scores
of Head Start students did not remain higher than those of
disadvantaged children who did not attend Head Start.
Finally, there is evidence to suggest that ideas that we
covered in the last discussion regarding socioeconomic status
are related to education variables, which together are
related to intelligence. Socioeconomic status might influence
aspects of education, and then intelligence, in a number
of ways:
Families with a high socioeconomic status often are able
to prepare their children for school because they have
access to resources (e.g., childcare, books and toys) and
information (what are the best schools? what aspects are
taught to children?) that enhance children’s social,
emotional and cognitive development and help parents
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 355
to better prepare their young children for school
(Demarest, Reisner, Anderson, Humphrey, Farquhar
and Stein, 1993).
Families of a low economic status face hard challenges
when it comes to providing the best care and education
for their children. When basic necessities such as money
and time are missing, food, housing, clothing and
healthcare come first. Educational toys and books, and
time searching out the best schools, are luxuries that
parents may not have the time or money to pursue
(Ramey and Ramey, 1994).
Parents from poor socioeconomic backgrounds often
grew up in poor socioeconomic conditions themselves.
Consequently, parents may have inadequate reading
skills or may lack knowledge about childhood nutrition
(or not be able to afford it), and these benefits aren’t
passed on to the children before school (Zill, Collins,
West and Hausken, 1995).
Culture and intelligence
The final environmental area that Neisser et al. (1996) suggest
is important to influencing intelligence is the cultural
environment people live in.‘Culture’ refers to people’s individual
values, and the values of their society. Nessier et al.
suggest culture can have an effect not only on intelligence
but also on the type of intelligence that might develop.
You remember that in Chapter 11 of this book, we outlined
implicit theories of intelligence. Those theories
showed how the definition of what constitutes intelligence
shifts across cultures, particularly when you compare Western
and Eastern cultures, and how conceptions of intelligence
shift across age and through different disciplines.
Well, clearly that discussion of implicit theories of intelligence
is relevant here (you may want to reread some of that
chapter).
However, in addition to implicit theories of intelligence,
Serpell (2001) identifies three concepts that explain how
intelligence in Western societies is set apart from other cultures
in the world. These three concepts are decontextualisation,
quantification and biologisation.
Decontextualisation
A lot of Western thinking is inherited from 3,000 years of
classical Greek philosophy. Socrates, Plato and Aristotle
were all Greek philosophers who have had a profound impact
on the way we think in our culture. In Western culture
there is a tendency to emphasise mathematics, the scientific
method and language. We make clear distinctions between
what is right and wrong, what constitutes justice, the need to
follow a logical progression and the idea that there are
higher and lower planes of ideas and activities – and perhaps
a universal truth. Decontextualisation is the ability
to disconnect, or detach oneself, from a particular situation
and think abstractly, and then generalise about it. Serpell
(2001) argues that the ability to think abstractly and generalise
about things has gained importance in Western society
because of industrialisation. With the growth of
capitalism there is a need for efficiency, some level of bureaucracy
and functionality (the ability to come up with
abstract principles) to help govern Western life, including
the markets (financial, housing, consumer), industry and
education. These needs become increasingly important.
Serpell questions the need to always view decontextualisation
as a sign of intelligence, and a failure to decontextualise
as a sign of unintelligence.
Quantification
Quantification is the act of discovering, or expressing, the
quantity of something. Serpell suggests that the study of
intelligence is surrounded by quantification in three ways:
First is the way intelligence theory and research is designed
to quantify intelligence. That is, when we
ask ‘what is intelligence?’ we are trying to encapsulate
a number of meanings and ideas into one word,
‘intelligence’.
Second is a concept called reification (see also Gould,
1981). Reification is the tendency to regard an abstract
idea as if it had concrete or material existence. We see intelligence
as something that is located in the brain, but
we do not know where it is located. Intelligence isn’t just
about certain processes; but it encapsulates things that
are not measurable, such as beauty or sophistication. For
many, the invention of the steam engine, or the wheel,
are highly intelligent acts; but they are also acts of beauty
to some, and they are certainly of sophistication to many.
Third is a tendency to quantify intelligence in terms of
numbers. We give people intelligence tests, which contain
an optimum number of items that should be completed
within a certain amount of time. These items
have been selected in accordance with studies that have
used statistical procedures to determine what aspects of
intelligence are out there. The test that is given is determined
by the person’s age. When participants have finished,
they are given scores for their performance, and
these scores are then transformed into IQ scores. These
IQ scores can then be compared against standardised
scores for the tests of people of the same age.
Serpell suggests that consideration of these three points indicates
that our understanding of intelligence is surrounded
by quantification. All this is not to say that
quantification is not a good process to understand intelli-
356 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 356
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 357
gence. To produce meaning and conceptual understanding
of words; to try to define, measure and locate concepts; and
to use numbers as objective criteria are excellent methods
by which to ensure the progress and understanding of intelligence.
However, Serpell suggests that when we are dealing
with something like intelligence, we must be sure that
we do not seek to over-quantify this concept.
Biologisation
Serpell (2001) raises our awareness that biological and evolutionary
theories have grown in prominence in late twentieth
century and early twenty-first century thinking
(biologisation). You will be aware that, in our psychophysiological
and evolutionary chapters (Chapters 8
and 9), we outlined the dramatic developments in the understanding
of genes, and that evolutionary psychologists
are able to link the evolution of the human species to animals
who lived billions of years ago. Many of the arguments
put forward by biological and evolutionary
psychology are convincing, and inspiring, in the understanding
of why we behave the way we do. However, Serpell
suggests some caution in overemphasising these models. It
must be remembered that many of these models talk about
developments over millions of years, and our advanced
study of ourselves is relatively new. Therefore, we must be
careful to ensure that our understanding of genetics, evolutions
over a long course of history, and genetic variation
are not used to explain intelligence within a relatively short
period of history.
Final comments on genetic
heritability and environmental
influences on intelligence
Perhaps the last word in this section goes to Bouchard and
Loehlin (2001), who suggest a framework for observing
sources of population variance in psychological traits (see
Figure 14.5). We presented this framework in Chapter 8
and think that it is applicable to intelligence.
To assess population variations in intelligence,
Bouchard and Loehlin’s framework is not only a good
overview of the debate but also sets some prudent criteria
in terms of assessing elements such as genetic effects and
environmental effects on intelligence. So, in all, we must
consider:
Genetic influences – including questions about what
gene is involved and what type of genetic variation (for
example, additive or non-additive). Is there a sex limitation
(e.g., brain size)?
Environmental influences – including, to what extent
does environment influence the genes, what types of environments
are involved (e.g., education, culture) and
are there gender effects?
Assortative Mating
Genetic and Environmental
Influence
Evolution
Intelligence
Genetic
1 To what extent is the
trait influenced by genes?
2 What type of gene is
involved?
3 How many loci are
involved?
4 Is there a sex limitation or
sex linkage?
5 Are chromosomal effects
involved?
Developmental
2 Do different environmental
factors influence during
development?
1 Do different genes
influence during
development?
Environmental
1 To what extent is intelligence
influenced by the environment?
2 What type of environment is
involved?
3 Are there gender effects?
4 Is transmission horizontal,
or is it vertical?
1 Are there any genetic x
environmental interactions
on intelligence?
2 What type are the
interactions between genes
and the environment?
1 What sort of selective factors
were at work during the
original evolution of
intelligence?
2 Are there current selective
factors at work?
3 Is intelligence an adaptation?
1 Is assortative mating present
in intelligence?
2 Are there sex differences in mate
preference for the intelligence?
Figure 14.5 Framework and questions regarding sources of population variance in intelligence.
Source: Based on Bouchard and Loehlin (2001).
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 357
Interaction between genetic and environmental
influence – including what type are the interactions
between genes and the environment (e.g.,nutrition,prenatal
causes)?
Developmental influences – including do different
genes influence during development, and do different
environmental factors influence during development?
Assortative mating – including is assortative mating
present in intelligence, and are there sex differences in
mating preference for intelligence?
Evolution – including what sort of selective factors were
at work during the original evolution of intelligence (e.g.,
different need for intelligences across different areas of
the world)? Are there current selective factors at work?
Clearly,when it comes to intelligence,some of these areas are
easier to identify or consider than others. For example, no
one has discovered whether there is a gene for intelligence;
nor can we be certain about what the different evolutional
demands on intelligence are. However, many of the areas –
assessing the level of genetic influence, the types of environmental
influence, and the possible interactions between
genes and the environment – are known, or at least provide
sources of evidence which provide evidence for a debate.Applying
Bouchard and Loehlin’s model to intelligence provides
a focus to an area that debates the relative influences of
(1) genes, (2) the environment and (3) the interactions between
genes and the environment on intelligence.
Section B – The bell curve:
race differences in intelligence
In this section we will outline what has become known as
the bell curve debate. Here, we consider evidence and arguments
concerning race differences in intelligence.
The bell curve
We have deliberately held back on discussing this material
until now, although it covers some of the topics we have
discussed in this and other chapters, because we wanted
you to be prepared for this discussion. In psychology, the
discussion of differences in intelligence, particularly race
differences, raises a lot of emotions and debate. We will,
later in this chapter, highlight this area of psychology further,
in the area of eugenics. We would like to think that
your reading of this material also raises emotions, not only
as a psychologist but also as a human being. However, we
would also like to think that you can see past some of your
emotions, pick out the argument and debate that exist
here, and use argument, rather than emotion, to decide on
the relative merits that are put forward (if you’re worried
about seeing this distinction, read more about academic
argument and fallacies in online Chapter 26 of this book).
The bell curve: intelligence and class
structure in American life
In 1994 two authors from the United States, Richard J.
Herrnstein and Charles Murray, published a book called
The Bell Curve: Intelligence and Class Structure in American
Life (Herrnstein and Murray, 1994). The book is an analysis
of IQ test scores in the United States. The term ‘bell curve’
refers to the shape of the distribution of a large number of
IQ scores in the United States – it looks like a bell (see
Figure 14.6).
The book caused huge debate because it reported many
things about intelligence, particularly the extent to which intelligence
is genetically inherited. It claimed to describe the
rise of a ‘cognitive elite’ in the United States – a social group
of persons with high intelligence, with an increasingly good
358 PART 2 INTELLIGENCE
Stop and think
Smoking and IQ scores
In 2005, Lawrence Whalley of the University of Aberdeen
and his colleagues (Whalley, Fox, Deary and Starr, 2005)
investigated smoking as a possible risk for intelligence
decline from age 11 to 64. In 1931, the Mental Survey
Committee in Scotland met and decided (due to there
being no reliable way of getting a representative sample)
to measure IQ scores for everyone in Scotland. So, on
Wednesday, 1 June 1932, nearly every child attending
school in Scotland who was born in 1921 took the same
intelligence test (n ϭ 89,498); this exercise was repeated
in 1947, testing almost all people born in 1936 (n
ϭ 70,805). Whalley et al. analysed a subsample of these
respondents between 2000 and 2002, looking not only
at smoking and IQ scores but also at childhood IQ
scores, level of education, occupational status, presence
of heart disease, lung function and hypertension.
Whalley et al. found that current smokers and nonsmokers
had significantly different IQ scores at age 64
and that differences remained after accounting for childhood
IQ score. All in all, smoking appeared to predict a
drop in IQ by just under 1 per cent.
What do you think are possible explanations for
smoking contributing to a fall in IQ scores?
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 358
%ofpopulation
25
0 100 170
IQ score
5
0 100 170
IQ score
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 359
Profiles
Richard Herrnstein and Charles Murray
Richard Herrnstein (1930–1994) was a prominent researcher
in comparative psychology. He was Charles
Sanders Pierce Professor of Psychology at Harvard University.
During his career, he worked with B. F. Skinner
(advocate of behaviourism, which seeks to understand
behaviour as a function of environmental histories of
reinforcement) in the Harvard pigeon lab, where he did
research on choice and other topics in behavioural
psychology.
In the 1960s, as part of this work on pigeon intelligence,
Herrnstein formulated the ‘matching law’, which
outlined how reinforcement and behaviour are linked.
The matching law addresses the idea of choice. It views
choice not as a single event or an internal process of the
person, but as a rate of observable external events over
time. Some have suggested that the matching law is an
important explanatory account of choice behaviour, and
its applications spread from psychology into other
fields, notably economics.
Charles Murray was born in 1943 in Newton, Iowa, in
the United States. He studied for his undergraduate degree
in history at Harvard. In 1965 he immediately joined
the Peace Corps for five years in rural Thailand (Peace
Corps volunteers travel overseas and work with people
there). By 1974 Murray had returned to the United States,
where he completed his PhD in political science at the
Massachusetts Institute of Technology. He has written
extensively on economics, crime and poverty and has become
one of the nation’s most influential thinkers. His
book Losing Ground: American Social Policy, 1950–1980
(Murray, 1984), was used by President Ronald Reagan
in developing his domestic policy. Murray has been the
subject of articles in Newsweek, The New York Times
Magazine and The Los Angeles Times Magazine.
chance of succeeding in life. This book noted several cultural
differences in intelligence, but its authors also made some
suggestions regarding the intellectual inferiority of certain
cultural groups. Such work was perceived by many as difficult
for US society. The authors’findings were seen as having
implications for social and public policy as well as pointing
to potential sources of inequality in the United States. We
will now briefly describe the main arguments put forward by
Herrnstein and Murray.
Herrnstein and Murray’s main arguments are built on
six premises (or as they present them, conclusions) about
intelligence. We have used Herrnstein and Murray’s original
words here, but have put simpler wording in parentheses:
1 ‘There is such a thing as a general factor of cognitive
ability on which human beings differ.’ (There exists a
general factor of intelligence, and individuals differ in
their intelligence; i.e., high, low, average intelligence.)
2 ‘All standardised tests of academic aptitude, or achievement,
measure this general factor to some degree, but
IQ tests, expressly designed for that purpose, measure it
most accurately.’ (There are many tests of academic aptitude
or achievement, but IQ tests, which are specifically
designed to measure intelligence, measure intelligence
most accurately).
3 ‘IQ scores match, to a first degree, whatever it is that
people mean when they use the word intelligent or
smart in ordinary language.’ (IQ scores reflect whatever
most people mean by the word intelligent.)
4 ‘IQ scores are stable, although not perfectly so, over
much of a person’s life.’ (Throughout an individual’s life,
their IQ score remains relatively stable.)
5 ‘Properly administered IQ tests are not demonstrably
biased against social, economic, ethnic, or racial
groups.’ (IQ tests, when they are properly administered,
are not biased against any social, economic, ethnic or
racial groups.)
6 ‘Cognitive ability is substantially heritable, apparently
no less than 40 per cent and no more than 80 per cent.’
(Genetic heritability of intelligence is between 40 and
80 per cent.)
Herrnstein and Murray then go on to discuss four main
ideas in their book: (1) the cognitive elite, (2) socioeconomic
variables and IQ scores, (3) the relationship between
race and intelligence and (4) the implications for social
policy.
Figure 14.6 The distribution of IQ scores: A bell
curve.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 359
IQ
Social and economic indicators Less than 75 75–90 90–110 110–125 Greater than 125
The cognitive elite: looking at the higher
end of the bell curve
In their book, Herrnstein and Murray begin by looking at
the high end of the bell curve – that is, at those who score
higher on intelligence tests – and examine the relationship
between intelligence and education. Herrnstein and
Murray analysed IQ scores and admissions to universities
and colleges over 50 years, and they found that the most
important factor in college attendance was the intelligence
of the students, not social class or wealth. It is this
fact, Herrnstein and Murray maintain, that places the
most intelligent people in US society in the same place –
university and colleges.
Herrnstein and Murray argue that the next stage of this
process is the relationship between university and colleges
and the workplace.University and college graduates are often
placed within a select few occupations, reserved for those
obtaining college and university degrees including teaching,
engineering, law, research, medicine and accounting.
The authors then draw on studies showing that intelligence
is related to efficient and proficient (advanced degree
of competence) employees. You may remember the finding
of two US psychologists, John E. Hunter and R. F. Hunter
(1984), who did a meta-analysis (a meta-analysis is a technique
that combines the results of several studies) that put
together the results of studies examining various predictors
at the start of a job with eventual job performance. They
found that the correlation between intelligence and job performance
was much larger than job performance and the
individual’s curriculum vitae, their previous experience, the
job interview and education (r ϭ 0.10). With findings like
this, Herrnstein and Murray conclude that intelligence is
the single most powerful predictor of job success and
workplace productivity.
Herrnstein and Murray suggest that there is the emergence
in the United States of the cognitive elite, based on a
separation of society through education (university and
colleges) and the workplace (certain professions such as
teaching, research, law and medicine), in which the central
factor is intelligence. Consequently, Herrnstein and Murray
predict that intelligence will soon become the basis of the
American class system, and people with higher intelligence
will be at the top of this class system.
IQ scores and social and economic
problems: looking at the lower end
of the bell curve
Next, Herrnstein and Murray turn their attention to the
lower end of the bell curve – that is, people with low scores
on intelligence tests. What Herrnstein and Murray do is to
look at several factors in the context of IQ scores; these factors
include poverty, schooling, unemployment, family life,
welfare dependency and crime.
An example of the type of analysis they performed is
given in Table 14.1. This table is adapted from Linda
Gottfredson (1997), who summarised Herrnstein and
Murray’s (1994) analysis, and we have picked out just a few
of their findings to illustrate their analysis. The first row of
the table shows the percentage of people who fall within
the range of standardised IQ scores (remember that IQ
360 PART 2 INTELLIGENCE
Table 14.1 Social and economic indicators broken down by sample, and divided in terms of IQ score category
and expressed as a percentage of the overall population.
US population distribution
Men who are unemployed for
more than 1 month out of year
People who are divorced within
5 years of marriage
Mothers with children with IQ
less than 75
Lives in poverty
Ever been in prison
Mothers in receipt of long-term
welfare support
Dropping out from high school
(secondary education)
5% 20% 50% 20% 5%
12% 10% 7% 7% 2%
21% 22% 23% 15% 9%
39% 17% 6% 7% 0%
30% 16% 6% 3% 2%
7% 7% 3% 1% 0%
31% 17% 8% 2% 0%
55% 35% 6% 0.4% 0%
Source: Reprinted from Gottfredson, L. S. (1997). Why g matters: the complexity of everyday life. Intelligence, Vol. 24, pp. 79–132. Copyright © 1997,
reproduced with permission from Elsevier.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 360
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 361
scores are transformed to a mean of 100). This standardisation
into a normal distribution is consistent with the practice
of standardising scores into a mean of 100 and a
standard deviation of 15 (there is more on this process in
Chapter 12 if you need to refresh your memory). Therefore,
the greater percentages of people are concentrated in the
middle, with lower percentages at either end of the curve.As
shown in the first row of numbers in Table 14.1, 50 per cent
of people fall within the middle category of 90–110 IQ
points range, with 20 per cent of individuals falling within
the 75–90 and 110–125 IQ points range, and 5 per cent of
people scoring less than 75 and greater than 125 IQ points.
Herrnstein and Murray examined each sample (i.e.,
number of people in poverty, number of people in prison)
in terms of their IQ scores distribution and expressed this
as a percentage of the overall population (i.e., less than 75;
between 90 and 110 points). As you can see, larger percentages
of people (expressed in terms of the overall population)
in each of the social and economic subsamples are
concentrated towards the lower end of the IQ curve, that is,
less than 75 IQ points and 75–90 IQ points.
Herrnstein and Murray used these statistics to develop
their argument. For Herrnstein and Murray the statistics
on poverty, schooling, unemployment and crime clearly
show that intelligence underlies many of these social and
economic problems. They suggest that low IQ scores are a
strong precursor to poverty, more so than any socioeconomic
conditions. With schooling, low IQ scores predict
whether people are likely to drop out of high school (secondary
education) and decrease the likelihood of a person
gaining a degree. The authors point out that the figures
suggest that a person with a high IQ score is likely to finish
high school regardless of their social or economic circumstances.
Herrnstein and Murray find that low IQ scores are
related to unemployment and are seen in people who are
injured more often or who have given up work. They find
that the best predictor of unemployment among men is
not socioeconomic status or education, but IQ scores. The
authors conclude that low IQ scores are associated with
higher rates of divorce, lower rates of marriage and higher
rates of illegitimate births. They also find that IQ scores are
related to family structure, with families comprising a nuclear
or traditional structure (2 parents, 2 children) scoring
higher on intelligence tests than did families with extended
members living in the family home. They suggest that the
disappearance of the traditional nuclear family in the
United States was a result of low intelligence. Additionally,
Herrnstein and Murray point to the fact that the average
US criminal has an IQ score about 8 IQ points lower than
that of the average US citizen, suggesting intelligence is
linked to a propensity for crime in the United States.
In other areas, Herrnstein and Murray are less certain,
but also suggest that long-term welfare dependency and
parenting are related to IQ scores. They argue that low
intelligence is the primary factor in predicting people’s first
use of welfare, but then a culture of dependency (learnt
from childhood experiences of parents and relatives being
on welfare) may emerge that leads to individuals using welfare
in the long term. Finally, Herrnstein and Murray point
out that low IQ scores among mothers correlate with their
children’s poor motor skills and with social development
and behavioural problems from four years and upwards.
Herrnstein and Murray point out that although being of
lower intelligence does not prevent individuals from being
good parents, family environments where damage is done
to a child’s intellectual development occur when parents
are on the lower end of the intelligence distribution.
In considering the relationships between these factors
and intelligence, Herrnstein and Murray build an argument
where they suggest that an individual’s intelligence is
more important than their socioeconomic status in predicting
their eventual economic and social welfare.
The relationship between race and IQ:
implications for social policy
Herrnstein and Murray next turn their attention to the differences
that occur between various ethnic groups. They
examine the different fertility patterns of groups on the intelligence
distribution, but generally they adopt the position
that intelligence is largely genetically heritable.
Herrnstein and Murray compared IQ test scores by
ethnicity. For example, they found that individuals of Asian
and Asian American ethnic origin scored on average
around 5 IQ points higher than white Americans, particularly
on measures of crystallised (verbal) intelligence. However,
it is in the comparison between white Americans and
black Americans that larger differences emerge.
On the Wechsler Intelligence test (remember this is a
measure of verbal and non-verbal [crystallised/fluid] intelligence),
Herrnstein and Murray reported that white Americans
score 15 IQ points higher than black Americans. On
the Stanford-Binet test (remember this is also a measure of
verbal/non-verbal intelligence), Herrnstein and Murray
reported that white Americans scored 18 IQ points higher
than black Americans. Therefore, on both intelligence tests,
the gap between the two races is almost 1 standard deviation
(15 IQ points). In terms of normed IQ scores, white
Americans averaged 102 IQ points, while black Americans
averaged 87 IQ points. Herrnstein and Murray also note
that the average IQ scores of immigrants coming into the
United States were 95 IQ points, lower than the national IQ
average (see Figure 14.7).
It is here that Herrnstein and Murray move towards the
most controversial part of the book. They say the evidence
suggests that certain social factors in the United States
are pushing down its citizens’ intelligence. These factors
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 361
include the number of children produced, which is greater
for women who have lower IQ scores than for those with
higher IQ scores; consequently, the United States is producing
more and more children born into lower-intelligence
environments. They suggest that as more and more immigrants
come into the country, they show lower levels of intelligence.
They suggest, then, that these factors are causing
a downward pressure on intelligence in the United States;
and as the average intelligence of the nation falls, social
problems such as poverty, unemployment and crime will
increase as these social problems are associated with people
with low intelligence scores. Additionally inherent in
Herrnstein and Murray’s argument is that, at least by implication,
those groups of people in the country with lower
intelligence (for example, immigrants, African Americans),
are potentially part of these social problems.
Moreover, Herrnstein and Murray suggest that the way
forward is not to address social problems by trying to compensate
for individual differences in intelligence through
supporting and increasing these people’s intelligence. They
point to previous attempts to raise intelligence, such as nutrition
(as mentioned earlier and in the last chapter) or
preschool programs (such as Head Start) that have not
been successful in permanently raising intelligence. They
also suggest that as intelligence has a 40 to 80 per cent genetic
heritability, there should be little expectation that
intelligence should rise in these situations.
Furthermore, they argue against affirmative action programmes
in the United States. Affirmative action (or positive
discrimination) is a policy providing advantages for
people of a minority group who are seen to have traditionally
been discriminated against. The aim of affirmative
action is to create a more equal society. Herrnstein and
Murray suggest that preferential access to education or
employment for African Americans and immigrant groups
(where institutions increase the selection and promotion of
candidates from these ethnic minority groups in the
United States) has also failed. The authors argue that such
approaches have not only led to a decrease of intelligence in
education and the workplace (leading to further downward
pressure on intelligence in US society and, therefore, increasing
the potential of social problems) but also caused
racial tension in education and the workplace. This tension
arises from resentment between ethnic groups due to lower
levels of student attainment – or more dropping out in education
– and poorer job performance in the workplace as
a result of people with lower intelligence undertaking study
or jobs to which they are not suited.
Herrnstein and Murray’s position is that American education
and the workplace have been ‘dumbed down’ as they
take on people with lower intelligence. The authors suggest
that while US education and work institutions have been
good in helping the underprivileged, this success has been
at the expense of gifted students. Thus, by being ignored,
gifted students have not been able to develop their true potential.
Herrnstein and Murray argue that their aim is to
point to places of inequality; they observe that pretending
inequality does not exist has not been sensible. Instead,
they suggest, we should try living with inequality; doing so
is preferable because it is reality.
Herrnstein and Murray suggest that the money for education
should be shifted from supporting disadvantaged
programmes for those with a low intelligence (where there
is a predominance of African Americans and other ethnic
minorities) to programmes that support those with a high
intelligence (where there is a predominance of white Americans
and Asians). They argue that it is in promoting and
further valuing the gifted, and in raising the intelligence of
the nation, that the future of the United States lies. An educational
system that aids the gifted will value and raise intelligence
throughout society, particularly the workplace;
and social problems such as poverty, unemployment and
crime will be reduced.
Criticisms of The Bell Curve:
Intelligence and Class Structure
in American Life
When first published, the The Bell Curve received a great
deal of positive publicity, including cover stories in
Newsweek and articles in Time, The New York Times and the
Wall Street Journal. But there was a large amount of negative
response, particularly in the scientific community,
suggesting the book was oversimplified and had flawed
362 PART 2 INTELLIGENCE
Black
distribution
50 60 70 80 90 100
IQ
110 120 130 140 150
White
distribution
Figure 14.7 Distribution of IQ scores for white
Americans (average 102) and black Americans
(average 87 IQ points).
Source: Adapted with the permission of The Free Press, a Division
of Simon & Schuster, Inc., from The Bell Curve: Intelligence and
Class Structure in American Life by Richard J. Herrnstein and
Charles Murray. Copyright © 1994 by Richard J. Herrnstein
and Charles Murray. All rights reserved.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 362
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 363
Profile
Stephen Jay Gould
Stephen Jay Gould (1941–2002) was born and raised in
New York. He obtained his PhD at Columbia and served
as a member of the faculty at Harvard beginning in
1967. Gould’s expertise was varied; he was a palaeontologist
(a person who studies the forms of life existing
in prehistoric or geologic times), evolutionary biologist
and historian of science. At the end of his career he was
the Alexander Agassiz Professor of Zoology at Harvard.
In 1972 he worked with Niles Eldridge to develop the
evolutionary theory of punctuated equilibrium. In this
theory, Eldridge and Gould suggested that evolution
does not always represent long, drawn-out processes;
instead, evolutionary change may sometimes work in fits
and starts and can happen relatively rapidly, particularly
in times of environmental stress.
analysis. Much of this commentary was in response to the
authors’ discussion of race differences in intelligence. In reaction
to this publicity and controversy surrounding The
Bell Curve, the American Psychological Association (the
largest and most influential psychological society in the
world) established a special task force to publish an investigative
report on the research presented in the book as well
as an analysis of what is known and unknown about intelligence.
This task force was headed by Ulric Neisser, then at
Emory University in the United States. After an extended
consultative process, Neisser chose a range of academics
representing a broad range of expertise and opinion in the
literature of intelligence, including Thomas J. Bouchard Jr,
Stephen J. Ceci, Diane F. Halpern, John C. Loehlin and
Robert J. Sternberg (who are all mentioned in these chapters).
After analysing Herrnstein and Murray’s conclusions
regarding race and intelligence, the task force agreed that,
while large differences exist between the average IQ scores
of African Americans and white Americans, there is no definite
evidence suggesting these differences are genetic; they
may, in fact, be cultural.
More extensive critical considerations of Herrnstein and
Murray’s book can be found in the literature. However, in
the following discussion, we are going to highlight some of
the critiques that appeared at the time of The Bell Curve so
that you can develop a critical analysis of the arguments
contained within Herrnstein and Murray’s analysis.
There are three main areas of criticism of Herrnstein
and Murray’s book that we will now outline:
analysis of the assumptions used by Herrnstein and
Murray in their analysis (led by comments by Stephen
Jay Gould);
statistical and evidence-based problems in Herrnstein
and Murray’s analysis (led by comments by Leon J.
Kamin);
a darker side of psychology related to Herrnstein and
Murray’s analysis (led by comments by Stephen Jay
Gould and Leon J. Kamin).
Analysis of the assumptions used by
Herrnstein and Murray
Stephen Jay Gould (Gould, 1995), an academic at Harvard,
starts to dismantle some of the premises within Herrnstein
and Murray’s book to attack their arguments regarding
what should happen in the United States (for example, that
money for education be shifted from supporting disadvantaged
programmes for those with a low intelligence to programmes
that support those with a high intelligence).
Premises are the foundation of any argument. You can
read (online) Chapter 26 of this book for more details on the
importance of premises in arguments (though Gould used
the term ‘assumption’, in this example they can be treated to
mean the same thing). For example, imagine you criticise a
particular person as useless (your argument about them) because
they haven’t done a job you asked them to, and you assume
they have not done the job because they are lazy (your
premise). If you find out later that they haven’t done the job
because they had hurt themselves, then this makes your argument
invalid (that they are useless) because the premises
that form your argument (that the person is lazy) are wrong.
Well, in the same way, Gould begins to dismantle some
of the premises informing the arguments that Herrnstein
and Murray develop in their book about what should happen
in the United States. We will now look at each of the
premises that Herrnstein and Murray use – and Gould
seeks to attack (though we have supplemented them with
evidence and argument from more recent considerations
so as to bring the debate up to date).
Assumption 1: ‘There is such a thing as a
general factor of cognitive ability on which
human beings differ.’
In Chapter 12 we outlined Spearman’s theory of a general
factor of intelligence (‘g’). However, we know from our discussion
of intelligence in Chapters 12 and 13, that there are
M14_MALT2908_02_SE_C14.QXD 10/23/09 10:22 AM Page 363
difficulties in assuming that there is a general factor of
intelligence (g). These difficulties reflect the work of theorists
and researchers who:
Use factor analysis to identify general factors of intelligence
and suggest intelligence is better understood as
comprised in a hierarchy of intelligence abilities. For
example, see Carroll (intelligence in three strata) or Cattell
(intelligence comprised of two related but distinct
components, crystallised intelligence and fluid intelligence)
(Chapter 12).
Challenge, entirely, the notion of a general intelligence,
and suggest it is best represented by a number of different,
and separate intelligences. For example, see Sternberg’s
triarchic theory and Gardner’s multiple
intelligence theory (Chapter 12).
Emphasise cognitive psychology processes and argue
that intelligence is about not only general factors of abilities
but also cognitive processes. For example, see the
Kaufman’s ability tests and Das and Naglieri’s Cognitive
Assessment System (Chapter 13).
As we can see, the assumption that there is a general factor
of cognitive ability (intelligence) on which human
beings differ is open to debate.
Assumption 2: ‘All standardised tests of
academic aptitude, or achievement,
measure this general factor to some degree,
but IQ tests expressly designed for that
purpose measure it most accurately.’
If we accept the previous counterargument, that a general
factor of cognitive ability (intelligence) on which
human beings differ is ‘open to debate’, then the assumption
that the intelligence tests expressly designed for
measuring IQ can measure intelligence most accurately is
also open to debate, as such tests cannot measure the
many different definitions of intelligence proposed by
the different theorists.
We have seen that, even with established general intelligence
(IQ) tests (described in Chapters 12 and 13),
there are distinctions between psychometric-based intelligence
tests that measure verbal and non-verbal general
intelligence (e.g., the Wechsler tests and the StanfordBinet
IQ), psychometric based non-verbal intelligence
(e.g., the Raven’s Matrices) and cognitive-based intelligence
tests that measure intelligence abilities and intelligence
processes (Kaufman’s abilities tests and the
Cognitive Assessment System). Therefore, we can see
that this assumption is open to debate, and may be particularly
important because Herrnstein and Murray rely
heavily on psychometric IQ test scores to build their
arguments.
Assumption 3: ‘IQ scores match, to a first
degree, whatever it is that people mean
when they use the word intelligent or smart
in ordinary language.’
There is a problem with this assumption if we consider implicit
(or everyday) theories of intelligence (which we covered
in Chapter 17). Implicit theories of intelligence
specifically address what people mean when they use the
word ‘intelligent’ in ordinary language.
The first point to emerge from studies looking at implicit
theories of intelligence is the difference between Western
and Eastern cultures of intelligence.You will remember that
intelligence in Western cultures suggests an emphasis on the
speed of mental processing and the ability to gather, assimilate,
and sort information quickly and efficiently. In Eastern
cultures, these ideas do not fully apply; instead, intelligence
extends to social, historical and spiritual aspects of everyday
interactions, knowledge and problem solving. It would be
unfair to extend this observation to debate Herrnstein and
Murray’s Bell Curve assumption as they were talking about
only one Western culture, the United States. Nonetheless,
they did build some of their arguments on the fact that immigrants
coming into the United States are scoring lower
than the national IQ score average and exerting a downward
pressure on the intelligence of the United States. Might it be
that these immigrants are coming from other countries
where different types of intelligence are emphasised?
Other findings with studies of implicit theories of intelligence
do extend directly to Herrnstein and Murray’s argument.
For example, our conceptions of intelligence do
change across the age range, or across academic disciplines,
in Western cultures. Remember the study mentioned in
Chapter 12 by Canadian psychologist Prem Fry (Fry, 1984)
who found that intelligence at the three main stages of educational
development – primary school (5 to 11-year-olds),
secondary school (11 to 18-year-olds), and tertiary (college
and university) levels (18+-year-olds) – differed strongly.
Also remember Sternberg’s study examining implicit theories
of intelligence among art, business, philosophy and
physics professors (Sternberg, 1985). He found that conceptions
of intelligence across these experts differed greatly.
Therefore, Herrnstein and Murray’s assumption that intelligence
test scores match, to a first degree, whatever it is
that people mean when they use the word ‘intelligent’ in
ordinary language, is widely open to debate.
Assumption 4: ‘IQ (intelligence test) scores
are stable, although not perfectly so, over
much of a person’s life.’
The assumption rests on whether intelligence can substantially
change over a person’s life. This is still a major debate;
364 PART 2 INTELLIGENCE
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:04 PM Page 364
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 365
in the intelligence literature, it is referred to as the fixed
versus malleable intelligence debate. There are two areas to
consider in examining this assumption.
First is the view that intelligence does not substantially
change during one’s lifetime (that is, it remains fixed). As
you will remember from Chapter 13, some authors have reported
that IQ scores remain stable over long periods of
time; these findings support the view that intelligence is
generally fixed. Jones and Bayley (Jones and Bayley, 1941)
found that IQ scores of children at 18 years were positively
correlated with their IQ scores at 12 years (r ϭ 0.89) and
6 years (r ϭ 0.77). Deary, Whalley, Lemmon, Crawford and
Starr (2000) found that over a period of 60 years, IQ scores
were correlated high (at least r ϭ 0.63). Also, evidence that
educational schemes such as Head Start among disadvantaged
children (outlined in this chapter and Chapter 13), in
which initial gains in IQ scores were not sustained, supports
the view that intelligence is fixed.
However, the second area to consider points to evidence
suggesting that intelligence is malleable (does or can
change). We described, in this chapter, Ceci’s (1990, 1991)
meta-analysis of hundreds of studies. His results suggest
that there are many effects of schooling and education on
IQ scores, including the findings that children who attend
school regularly score higher on IQ tests (a rise of 2.7 IQ
points) and that delays in starting school cause intelligence
scores to drop (by 5 IQ points per year missed). We also described
Wahlsten’s (1997) mention of adoption studies in
France: an infant is moved from a family with a low socioeconomic
status to a family with a high socioeconomic status,
and the child’s IQ score improves by 12 to 16 points.
Finally, you will remember our discussion of the role of nutrition
in intelligence, particularly the role of vitamin supplements
on IQ scores among children. Reviews of these
studies suggest that in the majority of cases, there is a positive
effect on intelligence as the result of vitamin and mineral
supplements in at least some of the children on
non-verbal (fluid) measures of intelligence (Benton, 1991).
The question of whether intelligence is fixed or malleable
is open to debate. Consequently, there is some evidence
in this area to question the assumption that IQ
scores are stable, although not perfectly so, over much of a
person’s life.
Assumption 5: ‘Properly administered
intelligence (IQ) tests are not demonstrably
biased against social, economic, ethnic, or
racial groups.’
The first thing to acknowledge in this area is that it was
agreed by the APA task force (Neisser et al., 1996) that
properly administered intelligence tests are not demonstrably
biased against social, economic, ethnic, or racial
groups. The argument that they are biased is often put forward
by opponents, but this view that intelligence tests,
themselves, are biased against such groups may be misguided.
Rather, a more sensible position to adopt at this
stage is not that intelligence tests are biased against such
groups, but that the theory from which they are developed
may be biased. For example, there is a general factor of intelligence,
or the view that intelligence comprises only
speed of mental processing and the ability to gather, assimilate
and sort information quickly and efficiently, that is
strongly emphasised in Western culture.
Assumption 6: ‘Cognitive ability is
substantially heritable, apparently no
less than 40 per cent and no more than
80 per cent.’
As we noted earlier in this chapter, a number of sources, including
the APA task force (Neisser et al., 1996) agree that
cognitive ability is substantially heritable – apparently no
less than 40 per cent and no more than 80 per cent. As you
already know, there is some convincing evidence that intelligence,
to some extent, is genetically inherited. However, it
is worth noting that this margin, of 40 to 80 per cent, is
quite large.
Gould’s issue with this assumption is that heritability
can then be used to explain differences for intelligence
scores between ethnic groups. You remember that earlier
we outlined a fallacy in comparing populations and gave
the example of comparing death rates among a population
of army recruits and the death rates for the population of a
capital city from the same general population. We noted
that you have to be aware of several different factors determining
death rates between these two different populations.
Well, this fallacy also exists in Herrnstein and
Murray’s application of genetic heritability to explain differences
between black and white Americans (though
Herrnstein and Murray do acknowledge this problem in
their work). However, Gould argues that substantial heritability
of intelligence within a population group (e.g.,
white or black Americans) cannot be used to explain average
differences between groups (white Americans versus
black Americans).
To extend this argument, evidence from geneticists
could even ask whether race is a valid distinction to make
between populations. In terms of understanding differences
in intelligence between populations, such as race, one
important context to consider in genetic influence is the
amount of variance that is accounted for by race in the first
place. We all know the saying that the colour of one’s skin is
only skin deep.Well, as we also know, this is a scientific fact.
Typically, people looking at race differences make those
distinctions based on what they can see; this is called
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 365
366 PART 2 INTELLIGENCE
socially defined race. As R. S. Cooper and his colleagues
note, researchers tend to define race within broad concepts
of white, African, Asian, Chinese or Hispanic (Cooper
et al., 2000). However, these distinctions generalise across a
huge number of different cultures and languages. For example,
are white Americans the same as white individuals
from the United Kingdom, Italy and Australia?
Due to advanced techniques in understanding our physiology,
our understanding of race differences have dramatically
changed. Authors such as US evolutionary biologists
Masatosi Nei and Arun Roychoudhury (Nei and
Roychoudhury, 1982, 1993) and Joseph Graves (Graves,
2001) note that the genetic variance within a population is
around 10 times greater than genetic variance between
races. In fact, two people who are closely related in terms of
their ‘race’ may differ many more times in their genetic
makeup than will two people of two separate ‘races’. Today,
biologists suggest the criterion by which we determine
race, which is mainly based on skin colour, does not make
sense in genetic terms. There are virtually no expressed
genes (where the gene’s information is converted into a living
cell) that can be found in all the members of one race.
In fact Mei also suggests that within genetic research,
most of the genes that influence appearance are not
known. However, biologists have used the similarities between
our common proteins to map human populations.
Proteins are essential to the structure and function of all
living cells and viruses, and play structural and mechanical
roles in our genetic makeup. When scientists do this mapping,
they get a different genetic picture of the world to socially
defined race. For example, when geneticists have
mapped the human population in terms of our blood types
(remember, there are A, B, AB and O blood types), they
find that the English, Spaniards, Eskimos and Norwegians
cluster together; the Australians, Aborigines and Sicilians
cluster together; the Icelanders and Japanese cluster together
and the Swedes and the Ethiopians cluster together.
In all, these considerations suggest that the validity of
the arguments that Herrnstein and Murray develop from
intelligence having some genetic heritability, and from race
differences in IQ scores, is seriously questionable.
Gould’s analysis, in which we are asked to consider the
strength of Herrnstein and Murray’s premises, begins to
undermine some of their arguments. Though none of the
assumptions are entirely wrong, there clearly exists debate
around four or five of them. This suggests that the premises
on which Herrnstein and Murray base their arguments
and recommendations are less secure than
originally presented.
If you want to see the two arguments presented in the
final form, go to Stop and think: The arguments of
Herrnstein and Murray and Gould. (It will help you to read
Chapter 26, Academic Argument and Thinking which can
be found on the book website.)
Statistical and evidence-based
problems in The Bell Curve arguments
Leon J. Kamin (a professor of psychology at Northeastern
University in the United States) questions some of the statistical
thinking and research evidence used by Herrnstein
and Murray (Kamin, 1995).
You may remember this statement from your first statistical
classes in psychology:‘Correlation does not mean causation’.
It is important to remember, when reporting any
sort of correlation, not to infer immediately that one variable
causes another. Therefore, if we found a relationship
between optimism and depression, it would usually be the
case that we would not infer causation. Often you will hear
it is likely that the two variables influence each other (you
can read more on this in Chapter 26). It is this criterion
that Kamin applies to much of the evidence used by Herrnstein
and Murray. Kamin suggests that Herrnstein and
Murray, in a lot of their work, ignore that principle in interpreting
correlations.
He also suggests that in Herrnstein and Murray’s pursuit
of determining what factors may cause differences in
intelligence, and what factors are caused by differences in
intelligence, they begin to blur the fact that they are just
using correlational data. They end up ignoring the ‘correlation,
not causation’ rule. Kamin uses as an example
Herrnstein and Murray’s analysis of the relationship between
socioeconomic status and intelligence, which is
central to many of their arguments. In exploring this relationship,
Herrnstein and Murray use data from the
National Longitudinal Survey of Labor Market Experience
of Youth, in which over 12,000 children provided data on a
number of variables, including their socioeconomic status
and their intelligence. As part of their analysis, Herrnstein
and Murray conclude that the intelligence of a person is
more important than their socioeconomic status in predicting
their eventual economic and social welfare. However,
as we know from discussions in this chapter and
Chapter 13, socioeconomic status is bound up in a number
of factors – including nutrition, educational success and
poverty – that are related to intelligence. Kamin’s problem
with Herrnstein and Murray’s analysis is that they force
causal relationships on the variables when this approach is
not appropriate. They do so by treating both IQ test scores
and socioeconomic status as causal variables of eventual
economic and social welfare. They then remove socioeconomic
status as a factor, as it is not as powerful as intelligence
in predicting eventual economic and social welfare.
This hides the fact that intelligence and socioeconomic status
are intrinsically linked not only with each other, but
with economic and social welfare success. Kamin argues
that this type of analysis is not valid.
Kamin also raises some questions regarding the research
basis of Herrnstein and Murray’s arguments. His first
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 366
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 367
Stop and think
The arguments of Herrnstein and Murray and Gould
Let us take you through the two major arguments
presented around Herrnstein and Murray’s theories.
Also remember that when we use the term ‘true’ here,
we actually mean that ‘a majority of evidence suggests’,
because in psychology we deal with probability estimates
of findings. We cannot ever say anything is completely
true, proved or disproved (see Chapter 26 if you
want a full explanation of academic argument).
Argument 1: Herrnstein and Murray’s
argument
Major premises
If it is true there is such a thing as a general factor of
cognitive ability on which human beings differ;
and it is true that all standardised tests of academic
aptitude or achievement measure this general factor
to some degree, but IQ tests expressly designed for
that purpose measure it most accurately;
and it is true that IQ scores match, to a first degree,
whatever it is that people mean when they use the
word;
and it is true that IQ scores are stable, although not
perfectly so, over much of a person’s life;
and it is true that properly administered IQ tests are
not demonstrably biased against social, economic,
ethnic or racial groups;
and it is true that cognitive ability is substantially
heritable, apparently no less than 40 per cent and no
more than 80 per cent;
Minor premises
and if it is also true that one population group identified
by their ethnicity had an average IQ score lower
than another population group identified by their
ethnicity;
Conclusion
therefore, it must be true that the population group
who had the lower average IQ score must be less
intelligent than the population group who had the
higher average IQ score.
Argument 2: The counter argument based
on Gould’s analysis
Major premises
it is not neccesarily true there is such a thing as a general
factor of cognitive ability on which human beings
differ (because of multifactor theories of intelligence);
it is not neccesarily true that all standardised tests of
academic aptitude or achievement measure this
general factor to some degree, but IQ tests expressly
designed for that purpose measure it most accurately
(because general intelligence tests are very different
today and there is a real distinction between
psychometric and cognitive ability tests);
it is not neccesarily true that IQ scores match, to a
first degree, whatever it is that people mean when
they use the word (because implicit theories of intelligence
suggest conceptions of intelligence vary
between and within cultures);
it is not neccesarily true that IQ scores are stable, although
not perfectly so, over much of a person’s life
(because there is evidence that suggests intelligence
does or can change substantially);
it is true that properly administered IQ tests are not
demonstrably biased against social, economic, ethnic
or racial groups (mainly due to the conclusion
drawn by the APA Task Force);
it is true that cognitive ability is substantially heritable,
apparently no less than 40 per cent and no more
than 80 per cent:
Minor premises
and if it is also true that one population group identified
by their ethnicity had an average IQ score lower than
another population group identified by their ethnicity;
Conclusion
therefore, it can’t be necessarily true (due to faults with
four of the major premises) that the population group
who had the lower IQ score must be less intelligent
than the population group who had the higher IQ score.
Which argument would you support?
concern is with the validity of some of the measures used by
Herrnstein and Murray to assess key variables. One of these is
the variable mentioned earlier – socioeconomic status data
from the National Longitudinal Survey of Labor Market
Experience of Youth. The measurement of socioeconomic
status was achieved by getting students’ self-report of their
socioeconomic status. No check was made in this study on
whether students were accurate in their reporting of socioeconomic
status (for example, by asking their parents). Kamin
suggests that while such reports are not completely unreliable,
they do cast doubts on the reliability of Herrnstein and Murray’s
building of arguments and analyses based on this data.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 367
368 PART 2 INTELLIGENCE
Kamin also suggests that there are problems with some
of the research studies used to support some of Herrnstein
and Murray’s arguments. Kamin suggests that Herrnstein
and Murray rely too heavily on the work of certain academics
in the world of intelligence, particularly that of
Arthur Jensen and Richard Lynn. Kamin argues that there
are problems with Jensen’s and Lynn’s work. One such
piece of work is Herrnstein and Murray’s reliance on a review
paper by Lynn in which he examines the evidence for
the evolution of race differences in intelligence and discusses
the genetic components of intelligence as explaining
differences between ethnic groups (Lynn, 1991). In his review
paper, Lynn describes a publication by Ken Owen
(who is with the Human Sciences Research Council,
Pretoria, in South Africa) as the best single study of racial
differences in intelligence (Owen, 1989). The study compared
white, Indian and black pupils (including pupils
from a Zulu tribe) on an intelligence measure called the Junior
Aptitude Tests. In this study Owen did not assign IQ
scores to any of the groups he tested; he merely reported
test-score differences between groups. Owen also noted
that the lower scores of black children (which there were)
would be due to the poor knowledge of English among the
black participants. Also, Owen reported that the tests used
figures of items such as electrical appliances and microscopes,
which were unknown to Zulus. However, Kamin
notes that despite these reservations by Owen about his
data, Lynn constructed IQ scores from the data and based
many of his conclusions in this review on apparent differences
between white, Indian and black children in the study.
Kamin points to other questions that surround Lynn’s
(1991) review paper and are then reported by Herrnstein
and Murray. Lynn also refers to a finding by A. L. Pons, who
tested 1,011 Zambian copper miners on the Raven’s Matrices
(Pons, 1974). Pons presented his data orally at the 1974
26th Congress of the South African Psychological Association.
However, later D. H. Crawford-Nutt reported this
finding alongside some other data, when reporting on test
and item bias in the use of the Junior Aptitude Test as a
suitable measure of IQ among white, Indian and black
pupils (Crawford-Nutt, 1976). Pons had found that the
copper miners performed poorly on the Raven’s Matrices,
and Lynn used this finding in his review paper to support
racial differences in intelligence. Yet, Kamin points out that
in his paper, Crawford-Nutt also presented data among 228
black high school students in Soweto, South Africa, who
had scored slightly higher on the Raven’s Matrices when
compared to the norms for the test for white students of
the same age in the country. Yet, Lynn ignores this finding
in his review paper.
Kamin also points to problems owing to Herrnstein and
Murray’s citing some of Jensen’s work (Jensen, 1993b) regarding
race differences in reaction time (an indicator of
IQ). In this work, Jensen had tested black and white reaction
responses to stimuli. As you may remember, there are
different ways of measuring reaction time. Simply, reaction
time measures how quickly a participant responds to any
particularly stimulus. However, Jensen had also developed
a measure of ‘choice reaction time’, a more complex task,
which required participants to react to various stimuli presented
in a random order. Jensen found that black participants
did better than white participants (suggesting they
have higher intelligence), but the result was reported as an
inconsistency.
For Kamin, the fact that Herrnstein and Murray have
used problematic research evidence, as well as faulty statistical
thinking, to build their arguments, casts doubts on the
validity of their final arguments.
A darker side of psychology related to
Herrnstein and Murray’s analysis
Gould referred to the ‘ghosts’ of The Bell Curve. We noted
at the beginning of this chapter that Jensen referred to a
number of comments about Herrnstein and Murray’s
book, including it being described as ‘neo-Nazi’. What are
these authors referring to? Why did The Bell Curve book attract
such comments? The answer is that many thought
The Bell Curve was raising issues regarding eugenics.
Eugenics
To say the role of eugenics in psychology is a controversial
one is an understatement. It is something that is rarely
talked about in psychology undergraduate classes, but psychology
has a dark history.
Eugenics refers to a selection process within human
reproduction with the intent to create children with desirable
traits. Generally, eugenics policies are divided into two
aspects: positive and negative eugenics. Positive eugenics
refers to the practice of encouraging increased reproduction
in those who are seen as having superior traits (for example,
higher intelligence). Negative eugenics refers to the
practice of discouraging or eliminating reproduction in
those perceived to have poor hereditary traits (for example,
lower intelligence).
To look at one of the first developments of eugenics, we
must go back to where we first began in this chapter, with
Francis Galton. Galton, you will remember, became interested
in studying the variations in intelligence after reading
Charles Darwin’s The Origin of Species, which described
how all species evolve through a process of natural selection
and survival of the fittest. In 1865, Galton outlined eugenics
principles (Galton, 1865). He felt that human society
sought to protect the weak, and that these principles were at
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 368
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 369
odds with the process of natural selection that suggested the
survival of the fittest.
As Galton felt that intelligence was something that was
inherited through the genes, he argued that one could use
artificial selection to increase intelligence among humans.
He argued that such selection was needed as the lessintelligent
people were reproducing more than the intelligent
people, and this was causing the human race to
become weaker.
Galton’s theories were adopted by other psychologists.
One notable supporter was Lewis Terman (whom we
mention in Chapter 12), at one time a president of the
American Psychological Association, who had used the
Binet-Simon test among California schoolchildren and
later adapted it into the Stanford-Binet test. However,
Terman continued Galton’s ideas. In Genetic Studies of
Genius (Terman, 1925; Terman and Oden, 1947, 1959), he
argued that low intelligence was often found in SpanishIndian,
African American and Mexican families, and such
low intelligence was inherited. He argued that such groups
should be segregated into special classes and could only
ever be trained to be efficient workers. He also expressed
concerns that these families seemed to breed more than
white Americans did, and from a eugenics point of view
(the selection of certain traits), this presented a problem as
it was necessary to preserve the United States from low
intelligence.
By then, eugenics views had made their way into social
policy, politics and the law. By 1922 the process by which
human selection was to be initiated was being made abundantly
clear: sterilisation (the act of making a person infertile,
unable to reproduce). A member of the US House of
Representatives Committee on Immigration and Naturalization,
H. H. Laughlin, published the Model Eugenical
Sterilization Law. This bill formed the basis of state sterilisation
laws, and in it Laughlin listed the types of people
who were to be subjected to mandatory sterilisation. He included
the feeble minded (low intelligence), the insane, the
criminal, the epileptic, the blind, deformed and the dependent
(e.g., orphans, homeless).
In 1927, the US Supreme Court ruled in the case of Buck
v. Bell and supported a new legislative law in the state of
Virginia. The law concerned a 17-year-old woman named
Carrie Buck, who was a resident at the Virginia Colony for
the Epileptic and Feebleminded – an asylum home for
epileptics, the mentally retarded and the severely disabled.
Carrie had the IQ score of a 9-year-old, and her mother,
who also resided at the colony, had a mental age of less
than 8. Carrie Buck had given birth to a daughter who, at
1 year old, was given an infant IQ test and was found to
be less than normal. In response to this, the state of
Virginia wanted to have the child sterilised against her will.
The US Supreme Court ruled in favour of the enforced
sterilisation; and in writing up the decision, Justice Oliver
Wendell Holmes wrote, ‘three generations of imbeciles are
enough’. By that part of the twentieth century, 29 US states
had laws allowing the compulsory sterilisation of individuals
thought to be mentally retarded, alcoholic or ‘having a
criminal nature’. In 1945, information from the Journal of
the American Medical Association suggested that over
42,000 people were sterilised in the United States between
1941 and 1943.
But the United States was not alone. During the twentieth
century, Canada, Sweden, Australia, Norway, Finland,
Denmark and Switzerland all had various types of eugenics
programmes. These included promoting different birth
rates among populations, compulsory sterilisation, marriage
restrictions, birth control and immigration control.
However, it was in Germany that eugenics became a central
focus, with Hitler, the Nazi Socialist Party and the Second
World War.
In 1925, Adolf Hitler had published Mein Kampf (the
book outlines Hitler’s major ideas, including violent antiSemitism)
in which he identified the ‘mentally unworthy’
among the African race and suggested that they mustn’t be
allowed to perpetuate their race. In 1933, Adolf Hitler and
the Nazi Socialist Party set up the sterilisation law, which
was directly based on the Model Eugenical Sterilization
Law introduced by H. H. Laughlin in 1922. In Germany,
before the Second World War and between 1933 and 1939,
over 20,000 people were sterilised for being feeble minded.
Then the Holocaust occurred. The Holocaust was the attempt
to eradicate entirely particular target groups. By that
time, sterilisation of target groups such as the mentally retarded
had extended to the extermination of whole groups
of people and races including the Jews, the Poles, Russians,
Communists, homosexuals, the mentally ill, the disabled,
intelligentsia and political activists, Catholic and Protestant
clergy, some Africans and common criminals. During the
Second World War, Germany occupied a number of countries,
including France and part of the Soviet Union. In
1939, seven million Jews were killed in central and eastern
Europe, three million in Poland and over one million in the
Soviet Union.
Following the Second World War, and the experience of
Nazi Germany and Hitler with his aspirations for the perfect
race, eugenics fell into disrepute. The Nuremberg Trials
were trials of Nazis involved in the Second World War and
the Holocaust, and these trials revealed to the world the
Nazis’ genocidal practices. Clearly, governments couldn’t
condone those policies that had been advocated by Hitler,
and many re-examined their eugenics-based policies. In
1948, and in response to this gross abuse, the United
Nations affirmed that men and women of full age, without
any limitation due to race, nationality or religion, have the
right to marry and to found a family.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 369
370 PART 2 INTELLIGENCE
Eugenics and the bell curve
However, what has eugenics got to do with The Bell Curve?
Well, critics of the book, such as Gould, felt that some of
the issues raised by Herrnstein and Murray echoed eugenics
thoughts and practices. Their emphasis on singling out
people with low intelligence and segregating aspects of society,
their emphasis on genetic influence on intelligence
and their concerns about immigration brought up some issues
that surround eugenics. These sorts of concerns about
Herrnstein and Murray’s work were not helped, as Kamin
points out, by the fact that one main advocate of eugenics
in modern intelligence research is an academic on whose
evidence Herrnstein and Murray largely relied – Richard
Lynn.
Richard Lynn, known for his work on race and a firm
supporter of the genetic heritability of intelligence, wrote a
book on dysgenics (the biological study of the factors producing
degeneration of genes in offspring, especially of a
particular race or species), which argued that eugenicists
were right in their belief that modern populations have
been deteriorating genetically in respect to their intelligence
(Lynn, 1996). In a second book (Lynn, 2001), he considers
what measures could be taken to rectify the effects of
dysgenics and argues that genetic improvement is likely to
evolve when women use in vitro fertilisation (IVF; fertilisation
of an egg in the laboratory) to grow a number of embryos,
then have them genetically assessed and before
selecting those with genetically desirable qualities. However,
it is important to note that Lynn does not condone
Hitler’s actions of genocide. Instead, he sees eugenics as
purely a scientific pursuit to establish what is known about
genetic inheritance and that the aims and objectives of eugenics
should be open to scientific scrutiny.
It is also crucial to note that at no point do Herrnstein
and Murray discuss eugenics research or policy. However, it
is the discussion of a number of issues in Herrnstein and
Murray’s work that have been previously linked to eugenics
(low intelligence, genetic inheritability of intelligence, race
differences, their concerns about immigration past) that
Gould is pointing to when he mentions the ghosts of The
Bell Curve’s past.
Final comments
Phew! What a debate! That certainly may have got your
blood boiling, but please remember when discussing such
things in class and in essays that the arguments arising
from the discussion must generally be academically based,
not personally based (to read more on fallacies, go to the
online chapter, Chapter 26, Academic Argument and
Thinking). Notwithstanding the discussion, you should
now be able to outline:
what is meant by genetic heritability in intelligence;
the different dimensions of genetic and environmental
influences that are thought to impact on intelligence;
the main points of Herrnstein and Murray’s Bell Curve
analysis;
some criticisms of Herrnstein and Murray’s Bell Curve
analysis.
Stop and think
To tell the truth, the whole truth and nothing but the truth
Consider the following story. It is part of an obituary
written about Richard Herrnstein in 1994, after his
death, by Charles Murray (Murray, 1994). Think about
this statement with regard to reporting group differences
in intelligence.
About four years ago, shortly after Dick (Richard
Herrnstein) and I had begun to collaborate on a
new book about intelligence and social policy (The
Bell Curve), we were talking over a late-evening
Scotch at his home in Belmont, Mass(achusetts).
We had been musing about the warning shots the
prospective book had already drawn and the heavy
fire that was sure to come. The conversation began
to depress me, and I said, ‘Why the hell are we
doing this, anyway?’ Dick recalled the day when, as
a young man, he had been awarded tenure. It was
his dream fulfilled – a place in the university he so
loved, the chance to follow his research wherever it
took him, economic security. For Dick, being a
tenured professor at Harvard was not just the perfect
job, but also the perfect way to live his life. It
was too good to be true; there had to be a catch.
What’s my part of the bargain? He had asked himself.
‘And I figured it out,’ he said, looking at me
with that benign, gentle half-smile of his. ‘You have
to tell the truth.’ (Murray, 1994, p. 22)
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 370
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 371
Summary
Sir Francis Galton suggested that man’s natural abilities
are inherited under the same conditions as
physical features of the animal world that had been
described by Darwin. Galton suggested that intelligence
is passed down to children through heredity.
Heritability of intelligence is the estimated assessment
of the extent to which intelligence is passed
down from parents to children through their genes on
average across the population.
There are largely three types of study that you will
regularly see in the heritability of intelligence: family
studies, twin studies and adoption studies.
Heritability estimates of intelligence vary greatly,
ranging from an average of 40 per cent to an average
of 80 per cent.
There are four general issues surrounding genetic
heritability estimates: conceptions of heritability and
the environment, different types of genetic variance,
the representativeness of twin and adoption studies
and assortative mating.
We identify five main areas in which to consider environmental
effects on intelligence.
The first area is biological variables (nutrition, lead
and prenatal factors) and the maternal-effects
model.
The second area is the consideration of family environment
and shared and non-shared factors. Non-shared
environments consider within-family factors and outside-family
factors, including context-specific socialisation,
outside the home socialisation, transmission of
culture via group processes, group processes that
widen differences between social groups and group
processes that widen differences among individuals
within the group.
The third area is socioeconomic status variables that
also include consideration of birth order, family size
and intelligence.
The fourth and fifth areas are education and culture,
respectively, the latter comprising consideration of
factors such as decontextualisation, quantification
and biologisation.
In 1994 two US authors, Richard J. Herrnstein and
Charles Murray, published a book called The Bell
Curve: Intelligence and Class Structure in American
Life (Herrnstein and Murray, 1994). The term ‘bell
curve’ refers to the distribution of a large number of
IQ test results in the United States.
Herrnstein and Murray analyse the distribution of
scores, comparing the high end and low end of the
distribution of IQ scores and discussing the emergence
of the cognitive elite, social and economic
problems and the relationship between race and IQ.
There are huge criticisms of the arguments presented
by Herrnstein and Murray. First, these criticisms
centre on analysing some of the premises
used in the argument, including whether (1) there is
such a thing as a general factor of intelligence, (2) all
standardised tests of IQ measure intelligence accurately,
(3) IQ scores match what people mean by intelligence
and (4) IQ scores are stable, although not
perfectly so, over much of a person’s life.
Other criticisms of The Bell Curve focus on the consideration
of the use of statistics and research ethics
and its perceived link with eugenics ideas.
Connecting up
This chapter should be read after Chapters 12 and 13.
There are also some links to Chapter 8 (Biological Basis
of Personality I: Genetic Heritability of Personality and
Biological and Physiological Models of Personality),
which outlines to a greater extent some of the issues
surrounding heritability estimates.
Also Chapter 26, the supplementary chapter that we
have provided online may need to be read in order to
gain better insight into some of the elements of academic
argument.
M14_MALT2908_02_SE_C14.QXD 10/23/09 10:20 AM Page 371
372 PART 2 INTELLIGENCE
Critical thinking
Discussion questions
• Discuss the relative importance of genes and the environment
in determining IQ.
• Do differences in average IQ scores have any relationship
to, or infer anything about, the race of an individual?
• Discuss the view that without Herrnstein and Murray’s
work, the public today would know a lot less about
what factors influence intelligence.
• Critically examine the implications for schooling and
work, assuming that intelligence consists of one general
ability and is inherited.
• The second part of this chapter outlines a discussion
around a book written by Hernstein and Murray in
1994. In 2007, a researcher in intelligence found himself
at the centre of a controversy when he made comments
about differences in intelligence between
different parts of the world. As Hawkes (2007) notes,
this controversy seems to occur once a decade. There
are some online newspaper articles detailing this 2007
controversy (the web locations of these articles are
given below). Read these articles. What are the main issues
emerging from these articles? Why do you think
this controversy keep reoccurring?
– http://www.timesonline.co.uk/tol/news/uk/
science/article2687364.ece
– http://www.timesonline.co.uk/tol/news/uk/
article2677098.ece
– http://www.timesonline.co.uk/tol/news/uk/
science/article2687425.ece
– http://www.guardian.co.uk/science/2007/oct/
20/genetics
– http://www.guardian.co.uk/international/story/
0,,2193899,00.html
– http://news.bbc.co.uk/1/hi/sci/tech/7052416.stm
Essay questions
• Critically discuss the view that intelligence is largely genetically
inherited.
• Critically assess the view that genetics is more important
than the environment in determining intelligence.
• Critically discuss the view that there is a ‘cognitive elite’.
• How important are socioeconomic status and variables
in assessing the heritability estimates of intelligence?
• How do narrow and broad definitions of IQ heritability
differ? How has our understanding of genetics informed
the nature versus nurture debate on intelli-
gence?
• To what extent do racial group differences in intelligence
exist? Identify some important considerations relating
to these findings.
• The main problem with Herrnstein and Murray’s bell
curve is that there are major flaws in the premises of the
argument they present. Critically discuss.
Going further
Books
Herrnstein, R. J. and Murray, C. (1994). The Bell Curve.
New York: Free Press.
Jacoby, R. and Glauberman, N. (eds). The Bell Curve
Debate. New York: Times/Random House.
Plomin, R. (2004). Nature and nurture: An introduction
to human behavioral genetics. London: Wadsworth.
Plomin, R., DeFries, J. C., McClearn, G. E. and McGuffin,
P. (2000). Behavioral Genetics: A primer. London:
Freeman.
Journals
Zyphur, M. J. (2006). On the complexity of race.
American Psychologist, 61, 179–180. In this chapter we
explained that the analyses of race within mainstream
psychology usually relies on simple social categories,
but we also argued that there are problems with this
method as it bears little resemblance to those methods
used in the genetic analysis of race. If you want to read
more about this issue, then a good short starting point
is this article. American Psychologist is published by
American Psychological Association. Available online
via PsycARTICLES.
In this chapter and the last, we looked at the notion of
whether IQ and intelligence can change. A good article
summarising the main issues is Howe, J. A. (1998). Can
IQ change? The Psychologist, 11, 69–72. This is freely
available online. You can find The Psychologist on the
British Psychological Society Website (http://www.
bps.org.uk/).
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 372
CHAPTER 14 HERITABILITY AND SOCIALLY DEFINED RACE DIFFERENCES IN INTELLIGENCE 373
Maccoby, E. E. (2000). Parenting and its effects on children:
On reading and misreading behavior genetics.
Annual Review of Psychology, 51, 1–27. Annual Review of
Psychology is published by Annual Reviews, Palo Alto,
California.Available online via Business Source Premier.
Baker, L. D. and Daniels, D. (1990). Nonshared environmental
influences and personality differences in adult
twins. Journal of Personality and Social Psychology, 58,
103–110. Published by the American Psychological
Association. Available online via PsycARTICLES.
Bouchard, T. J. Jr and Loehlin, J. C. (2001). Genes,
personality, and evolution. Behavioural Genetics, 31,
243–273. Published by Kluwer Academic Publishers.
Available online via Kluwer or SwetsWise.
Harris, J. R. (1995). Where is the child’s environment?
A group socialization theory of development.
Psychological Review, 102, 458–489. Published by the
American Psychological Association. Available online
via PsycARTICLES.
If you are interested in looking deeper into the genetics
of brain structure and intelligence then a possible article
to read is Toga, A. W. and Thompson, P. M. (2005).
Genetics of brain structure and intelligence. Annual
Review of Neuroscience, 28, 1–23.
Articles on the intelligence issues discussed in these
chapters are often found in these journals.
Intelligence: A multidisciplinary journal. Published by
Elsevier Science. Available online via Science Direct.
American Psychologist. This journal has a number of
articles relating to the heritability of intelligence. Published
by the American Psychological Association. Available
online via PsycARTICLES.
Web links
In this chapter we have covered a number of topics
including heritability of intelligence, the bell curve and
birth order. Some of these topics are controversial and
evoke a lot of emotions, and therefore there are a lot of
websites that discuss these issues; but the evidence they use
is not appropriate. We would suggest that when it comes to
topics like these it is best to adhere to academic books and
journals. However, if you want to search the web, then the
place to start is the Hot Topics section of the Human
Intelligence website (http://www.indiana.edu/~intell/).
Film and literature
David Copperfield, by Charles Dickens (1869) and Under
the Greenwood Tree, by Thomas Hardy (1872). Our first
discussion of intelligence in this chapter was to do with
Galton’s view of intelligence and the reasons underlying it
at the beginning of the twentieth century. Galton looked
to members of the Royal Society and saw eminence as an
indicator of intelligence. However, the argument of what
surrounds intelligence was demonstrated in many novels
of the time. The writing of Charles Dickens (for example,
David Copperfield) and Thomas Hardy (Under the Greenwood
Tree) used intelligence to distinguish the main characters.In
both books,there is a contrast made between the
small village family, whose members are intelligent in
terms of the countryside around them and have inherited
wisdom passed through generations of their family, and
the city family,whose members are eminent,rich and own
businesses but are actually portrayed as incredibly unwise
compared to the village families. This treatment contrasts
greatly with Galton’s assumptions about the study of eminent
people living in London as reflecting intelligence.
Gattaca (1997, directed by Andrew Niccol). Vincent
(played by Ethan Hawke) is one of the last ‘natural’
babies born into a sterile, genetically enhanced world,
where a person’s life expectancy and the likelihood of
disease are ascertained at birth. Born with a heart defect
and due to die at 30, Vincent has no chance of a career
in a society that now discriminates against your
genes, instead of your gender, race or religion. This film
suggests a natural progression of eugenics, and raises
some issues about what happens to those who fall foul
of the perfect genetic world.
The Intelligent Man (1984, Educational Resources
Film). This video looks at the sometimes controversial
work of Sir Cyril Burt and his work with intelligence
tests and theory. The filmmaker looks at the experiments
and findings of the work carried out by Cyril
Burt and contrasts intelligence theories. BBC Videos
for Education and Training.
Explore the website accompanying this text at www.pearsoned.co.uk/maltby for further resources to help
you with your studies. These include multiple-choice questions, essay questions, weblinks and ideas for
advanced reading.
M14_MALT2908_02_SE_C14.QXD 10/21/09 4:05 PM Page 373