Chapter 6

Genetic and Environmental

Constraints on Variability
in Sport Performance
Joseph Baker, PhD and Keith Davids, PhD
Editors Overview
Why do some athletes benefit more than others from training and practice? A frequent observation of interindividual variation in response to training and practice
raises important theoretical and practical questions about the nature of genetic and
environmental constraints on skill acquisition and performance. This problem is a
manifestation of the long-standing debate of nature versus nurture, which argues
the precise proportion of performance variation in a population accounted for
by genetic characteristics or environmental influences. The relationship between
genetic and environmental constraints on responses to practice and training is
complex, requiring a careful interpretation of the data in the extant literature and
a comprehensive theoretical model to explain research findings. In this chapter, the theories of practice emphasizing the role of environmental constraints
in explaining variability in expertise are evaluated, and the evidence favoring
the role of genetic constraints in variability of interindividual responsiveness to
training and practice is examined. The biological determinism underlying some
recent interpretations of the roles of genetic diversity and environmental context
on variation in human motor performance is rejected for an interactive model that
is captured well by dynamic systems theory. The challenges for future research on
the interacting constraints of genetics and environment are (1) to locate the primary
and secondary influences on performance and (2) to understand the dimensions
of their interactions in order to improve practical intervention programs such as
those dedicated to talent identification and development.

hy do some athletes benefit more from training and practice? Are elite performance institutes justified in putting large amounts of funding into genetic
testing of athletes to micro-manage personalized training programs? (Dennis, 2005).
The frequent observation of interindividual variations in responsiveness to training and practice raises important theoretical and practical questions like these on

109

110

Baker and Davids

the genetic and environmental constraints on skill acquisition and performance.

How these constraints shape variations in performance is of increasing interest in
psychology, physical education, movement science, biology, and sports medicine,
and in this chapter we examine current theory and data on environmental and
genetic influences on expertise and performance. This issue is manifested in the
long-standing debate of nature versus nurture, which seeks to identify the precise
proportion in which genetic characteristics and environmental influences contribute
to variation in performance. Much has been written about this particular dualism,
and resolving the debate has proved difficult (for excellent analyses see Lewontin,
2000 and Johnston & Edwards, 2002).

The relationship between genetic and environmental constraints on responses

to practice and training is a complex issue that requires careful interpretation of the data in extant literature and a comprehensive theoretical model
to explain research findings. We begin this chapter by evaluating theories of
practice that emphasize environmental constraints in explaining variability
in achieving expertise. We then examine the evidence that favors the role of
genetic constraints on performance variability. We conclude by outlining the
theory of dynamic systems as a powerful explanatory framework for interpreting the interactional influences of genetic diversity and environmental context
on variation in human motor performance.

Nurture Perspective of Expertise Development:

Deliberate Practice
One of the most radical perspectives regarding the role of practice in performance variation is the framework of deliberate practice presented by Ericsson
and colleagues (Ericsson, Krampe, & Tesch-Rmer, 1993). They proposed that
individual differences in performance in any domain can be accounted for by the
amount and type of practice previously performed. Likewise, they suggested that
genetics play a minimal role in determining individual achievement and that this
role can be circumvented by optimal amounts of quality practice. While many
presumptions of this theory remain to be proven, deliberate practice is largely
based on two previously observed guidelines: the 10-year rule (Simon & Chase,
1973) and the power law of practice (A. Newell & Rosenbloom, 1981).

The 10-Year Rule

In their classic study of chess expertise, Simon and Chase (1973; Chase &
Simon, 1973) made the first suggestion that interindividual variation in performance can be explained by quantity and quality of training. This hypothesis was based on findings indicating that differences between expert (grand
master) and lower (master and novice) levels of skill were attributable to the
ability to organize information into more meaningful chunks rather than to a
superior memory. Since then, researchers examining experts and novices have

Genetic and Environmental Constraints on Variability in Sport Performance

found no reliable differences in static, physical capacities such as visual acuity,

reaction time, or memory (hardware) but have found consistent differences in
domain-specific strategies for information processing (software) (for a review
see Starkes, Helsen, & Jack, 2001). In a recent overview of the last 30 years
of research on expertise in sport, Singer and Janelle (1999) summarized the
characteristics that distinguish the expert:
1. Experts have greater task-specific knowledge (McPherson, 1993; McPherson & French, 1991).
2. Experts interpret greater meaning from available information (Abernethy,
1987; 1990; 1991).
3. Experts store and access information more effectively (McPherson,
1993).
4. Experts are better at detecting and recognizing structured patterns of
play (Allard & Starkes 1980; Chase & Simon, 1973).
5. Experts are better at using situational probability data (Abernethy &
Russell 1984; 1987).
6. Experts make decisions that are more rapid and appropriate (A.M. Williams, 2000).
In sport, research examining interindividual variation in cognitive abilities
has been somewhat limited to sports and physical activities with dynamic task
constraints demanding a high level of decision making. However, existing evidence suggests that in fields where the distinguishing characteristics between
experts and nonexperts are domain-specific abilities in information processing,
these differences result from training rather than innate ability. An interesting
question is the role of other genetic constraints, such as differences in power
or endurance, which we examine later in this chapter.
The 10-year rule stipulates that a 10-year commitment to high levels of
training is the minimum requirement to become an expert. This rule has
been retrospectively applied to the study of expert careers, with some success
in domains such as music (Ericsson, Krampe, & Tesch-Rmer, 1993; Hayes,
1981; Sosniak, 1985), mathematics (Gustin, 1985), swimming (Kalinowski,
1985), distance running (Wallingford, 1975), and tennis (Monsaas, 1985).
The perspective of deliberate practice (Ericsson et al., 1993) extends the work
of Simon and Chase (1973) by suggesting that it is not simply any training that
differentiates individual performance, but engagement in deliberate practice. By
definition, deliberate practice is not intrinsically motivating, it requires effort
and attention, and it does not lead to immediate social or financial rewards.
Further, involvement in deliberate practice depends on the learner accessing
effective resources (facilities, coaches, financial support), providing the necessary physical and mental intensity for progressively adapting to appropriate
training loads and possessing the ability to maintain involvement without

111

112

Baker and Davids

intrinsic forms of motivation such as enjoyment. In the framework of deliberate practice, future experts perform training that develops required skills under
continuously evolving conditions in which stress and recovery are optimally
balanced to maximize training adaptations and minimize training plateaus.

Power Law of Practice

Research examining the accumulated effects of prolonged practice and the rate
of learning has suggested that performance increases monotonically according to a power function. This finding, known as the power law of practice (or
the loglog linear learning law) (A. Newell & Rosenbloom, 1981), has been
demonstrated in numerous domains. According to the power law of practice,
learning occurs rapidly at the start of practice, but this rate of learning decreases
over time as practice continues (see figure 6.1).

FIGURE 6.1 Example of the power law of practice for performance on a response
time task.

Central to the notion of deliberate practice is the monotonic benefits assumption.

Ericsson et al. (1993) proposed that, contrary to the power law of practice, a
monotonic relationship exists between the number of hours of deliberate practice performed and the performance level achieved. Their original research with
musicians indicated that the difference between expert and nonexpert pianists
and violinists was the amount of time spent practicing while alone (i.e., deliberate practice). The best musicians had spent at least 10,000 h practicing alone
while their less successful counterparts had spent no more than 7,000 h.
Ericsson et al. further argued that it is not simply the accumulation of hours
of deliberate practice that leads to superior performance. The accumulation of
such hours must coincide with crucial biological and cognitive development.
Early specialization is an important element predisposing future success. Figure
6.2 illustrates the relationship of chronological age, time spent in deliberate
practice, and performance. Performers beginning deliberate practice at later
ages (performers b and c), even with the same commitment to training, are

Genetic and Environmental Constraints on Variability in Sport Performance

FIGURE 6.2 Relationship of chronological age, performance, and hours of deliberate practice.
Adapted from Ericsson, Krampe, and Tesch-Romer 1993.

unable to match the quantity of training accumulated by performers starting

earlier (performer a). The assumption that future experts must specialize early
becomes increasingly important in sports where peak performances typically
occur at younger ages (e.g., diving, gymnastics, and figure skating for women),
although the necessity for early specialization in sports where peak performance
occurs later (e.g., basketball, field hockey) has recently been questioned (Baker,
2003; Baker, Cote, & Abernethy, 2003a).

Deliberate Practice in Sport

Although the theory of deliberate practice was developed through research with
musicians, Ericsson and colleagues have indicated that it should also apply to
expertise in sport (Ericsson et al., 1993; Ericsson, 1996). To date, researchers
examining the theory of deliberate practice in sports have investigated figure
skating (Starkes, Deakin, Allard, Hodges, & Hayes, 1996), karate (Hodge &
Deakin, 1998), wrestling (Hodges & Starkes, 1996), soccer (Helsen, Starkes, &
Hodges, 1998), middle distance running (Young & Salmela, 2002), field hockey
(Baker et al., 2003a; Helsen et al., 1998), triathlon (Baker, Cote & Deakin,
2005; Hodges, Kerr, Starkes, Weir, & Nananidou, 2004), basketball (Baker et
al., 2003a), and netball (Baker et al., 2003a). These studies have encountered
some problems with applying the original framework of deliberate practice to
the sport domain. For example, Starkes and colleagues (Helsen et al., 1998;
Hodges & Starkes, 1996) found that athletes tended to rate relevant practice
activities as very enjoyable and intrinsically motivating, contrasting with a key
component of the definition of deliberate practice. Further, there is concern
regarding which forms of athletic training constitute deliberate practice. In the

113

114

Baker and Davids

original work of Ericsson et al. (1993), only practicing while alone met the
requirements for deliberate practice. In studies of deliberate practice in sport,
there are few, if any, training activities that meet the original criteria set in the
definition by Ericsson et al. (1993). Helsen et al. (1998) suggested that the
specifications for deliberate practice in sport should be extended to include all
relevant forms of training. This is particularly important in team sports where
both individual and team practices increase skill and improve performance.
The relationship between hours spent in practice and attainment is typically
consistent with the tenets of deliberate practice. Expert athletes accumulated
more hours of training than nonexperts (Helsen et al., 1998; Starkes et al., 1996;
Hodge & Deakin, 1998). Not only do experts spend more time in practice, but
they also devote more time to the specific activities deemed as being the most
relevant to developing the essential component skills for expert performance.
For example, Baker, Ct, and Abernethy (2003b) found that expert athletes
from basketball, netball, and field hockey accumulated significantly more hours
in video training, competition, organized team practices, and one-on-one coaching than nonexpert athletes.

Deliberate Practice and Interindividual Variation in Performance

The essence of the perspective of deliberate practice seems to be all individuals are created equal. In a review of studies on skill acquisition and learning,
Ericsson (1996) concluded that, with few exceptions, the level of performance
was determined by the amount of time spent performing a well defined task
with an appropriate difficulty level for the particular individual, informative
feedback, and opportunities for repetition and corrections of errors (p. 20-21).
Continually modifying the task difficulty allows future experts to perpetuate
adaptations to greater training stress. Informative feedback and opportunity
for repetition allow the performer to master skills more easily and to progress
more quickly.
Data from the Ericsson et al. (1993) study of expert musicians support the
relationship between hours of deliberate practice and level of performance.
Specifically, the study found that expert musicians spend in excess of 25 h/wk
in deliberate practice (training alone) whereas less successful musicians spend
considerably less time in deliberate practice (amateurs spend <2 h/wk). These
notable differences in weekly hours accumulate to enormous divisions after
years of training. Similar relationships have been found in chess (Charness,
Krampe, & Mayr, 1996). Prior research on the training histories of athletes and
the characteristics that distinguish individual athletes provides evidence for the
powerful role of appropriate training in building the expert sport performer.

Influence of Other Activities

Recently, researchers have provided evidence that challenges one of the basic
tenets of the theory of deliberate practice, specifically, that early specialization
is necessary for the development of expertise. Baker et al. (2003a) studied

Genetic and Environmental Constraints on Variability in Sport Performance

experts from field hockey, basketball, and netball and found that these players
performed a wide range of sports during early stages of development. As the
athletes developed, their broad involvement in sports gradually decreased until
they specialized in their main sport (figure 6.3). Moreover, Baker et al. (2003a)
reported a negative correlation between the number of other sports played and
the number of sport-specific training hours performers required before making
their respective national teams. These findings suggest that participation in indirectly related activities may augment the physical and cognitive skills necessary
for an athletes primary sport. For example, many of the athletes participated in
various forms of football (including rugby, Aussie rules, and touch football), a
sport that also requires dynamic, time-constrained decision making as well as
physical elements such as cardiovascular fitness and coordination.

FIGURE 6.3 The number of sporting activities performed each year by experts and
nonexperts from basketball, netball, and field hockey.
Adapted from Baker et al., 2003a.

Research by Ct (1999; Ct & Hay, 2002) has indicated that playlike

activities (termed deliberate play) during the early stages of training benefit
expertise development in many sports. Deliberate play represents the antithesis
of deliberate practice in that it is made up of activities designed for enjoyment that require active and pleasurable participation. In early development,
activities that are inherently enjoyable and motivating may be necessary to
provide an impetus to continue training when more diligent, effortful practice
is required. Without this pleasurable involvement, athletes may drop out of
sport (Petlichkoff, 1993).
The above relationships are not unexpected. During early stages of development, improvement comes rapidly and easily because there is so much room for
it. During this time, it is likely that any form of relevant participation provides
improvement, regardless of whether this participation is direct involvement
through sport-specific training or indirect involvement through sports that

115

116

Baker and Davids

share basic characteristics. However, as performance improves, enhancement

becomes increasingly difficult until focused training on specific areas of weakness becomes the only means of advancement. At this point, deliberate practice becomes the most effective form of training (see Baker, 2003 for a more
through review).

Challenges of Deliberate Practice

Despite the evidence favoring environmental effects on responses to practice and
training, the issues in the nature/nurture area are enormously complex. In her
thought-provoking paper on the theory of deliberate practice in sport, Starkes
(2000) raised some important questions regarding the relationship between
expertise and responsiveness to training and practice. Starkes examination
of the data on the elusive element of athletic success attempted to contrast
two theories purporting to explain athletic achievement: the sport commitment model of Scanlan and colleagues (Scanlan, Carpenter, Schmidt, Simons
& Keeler, 1993) and the theory of deliberate practice (Ericsson et al., 1993).
Despite the arguments proposed by these researchers, both theories end with the
feeling of the chicken or the egg. For example, the sport commitment model
provides few indications of whether the commitment required of performers of
international caliber is developed or is predominantly inherited. There seems
a hopeless correlation among the innate factors influencing the propensity to
enjoy sport and the willingness to invest time and effort into countless hours
of practice and the influence of positive learning experiences (which the coach
can do much to foster).

Deliberate Practice and Intraindividual Variability in Performance

K.M. Newell and McDonald (1991) have argued that practice is a necessary but
not sufficient condition for developing motor expertise. Traditional approaches
to practice have tended to overemphasise the amount of required time to be
spent in practice to the detriment of understanding how the quality of specific
practice activities affects expertise. For example, earlier in this chapter we
noted how the theory of deliberate practice is grounded in the power law of
learning. Despite the fact that the power law has been called the ubiquitous
law of learning (A. Newell & Rosenbloom, 1981, p. 2), there has been recent
criticism of this view, particularly as it relates to intraindividual variability in
performance.
K.M. Newell, Liu, and Mayer-Kress (2001) have pointed out that performance
can change in persistent and transitory ways as a function of learning and
development and that previous research has emphasized identifying persistent
changes rather than transitory changes. A number of variables can be used to
assess intraindividual change in performance over time, but task outcome has
been the main variable used in studying learning curves. The power law of

Genetic and Environmental Constraints on Variability in Sport Performance

learning has been the generally accepted law of learning primarily because data
from two of the best-known studies on learning, one by Crossman (1959) and
one by Snoddy (1926), fit the power law well. The power law of learning has
had some passionate advocates, including Logan (1988), for example, who
advocated that any theory providing data that do not fit the power law should
be immediately rejected.
But, as K.M. Newell et al. (2001) have indicated, careful analysis of the data
from Crossman (1959) (late in practice) and Snoddy (1926) (early in practice)
shows occasional significant departures from the power law. The notion that
there may be many functions of change over time supports a broad vision of the
factors that can influence change. K.M. Newell and colleagues (2001) criticized
two main methodological practices: (a) blocking trials and (b) averaging scores
over participants in studies of intraindividual change as a function of learning
and development. Trial blocks and averaging scores over individuals ignore the
fact that laws of learning should reflect both transitory and persistent changes,
whereas the power law treats transitory effects as random behavior, possibly
masking the persistent trends in intraindividual variations. Traditionally, this
behavior has been viewed as the result of noise or effects such as the warm-up
decrement (the result of early trials within a session that bring the performer back
up to the stable performance point reached in earlier practice). K.M. Newell et al.
(2001) admit that it is not clear to what extent averaging practices has affected the
data on learning, but future research clearly needs to consider how ubiquitous
the power law of learning actually is. Another point is that most experiments on
motor learning have been conducted with a span of a few hours at most. Using
a short duration for measuring learning and coming up with learning curves
naturally limits the range of curves that can be exhibited by learners. Hence there
is an inherent methodological bias that predisposes outcomes toward the curve
of the power law. That is, in real life several functions of change can emerge in
learning curves from multiple timescales of motor learning.
In contrast, the timescale of transitory change during learning is much shorter
than that of the persistent changes. But these variations should not be dismissed
as random or as the result of noise (K.M. Newell et al., 2001). Changes in the
outcome of action over time are the product of many interacting subsystems,
each with its own timescale that is continuously evolving over real time. Contrary to the power law edict that larger absolute gains in performance occur
early in learning before tailing off, the greatest absolute changes in learning
may occur any time during practice, particularly if the performer is learning
a new pattern of coordination. The longer a performer has been practicing,
the more likely there will be sudden discontinuous jumps in learning due
to developmental changes occurring over the life span. Exponential learning
curves are most likely to be found in the learning of simple motor tasks such
as the linear positioning and timing tasks of laboratories. This is because new
patterns of coordination do not need to be learned and qualitatively new patterns do not need to be picked up.

117

118

Baker and Davids

To summarize, this criticism of the traditional literature is reflected by the

assumption of monotonic benefits proposed by Ericsson and colleagues (1993).
The theory of deliberate practice places a narrow emphasis on the main constraints of time and effort spent practicing, regardless of how much innate ability
the performer brings to a learning situation or of the nature of the activities
that take place during practice (Ericsson, 1996; Ericsson & Charness, 1994;
Ericsson & Lehmann, 1996; Ericsson et al., 1993). Starkes (2000) argued that
this emphasis on a limited number of constraints leads to a very environmentalist theory., and the implication is that the constraints on skill acquisition
may range far wider than implied by the theory of deliberate practice (Davids,
2000). Sternberg (1996) has highlighted the fact that the current version of the
theory of deliberate practice ignores the issue of genetic constraints; is difficult
to disconfirm because its operational definitions are weak; has confounded
talent, motivation, and level of practice; and has no strong experimental work
involving control groups.
It is clear that acquiring expertise in sport takes time, although the time and
amount of deliberate practice should not be viewed as the only constraints on
skill acquisition. The overarching theoretical question is: In what mechanisms
does interindividual variability in performance manifest itself? For example,
the issues of hereditary influences, such as baseline differences in motivation,
should not be ignored merely because the work in this area is ongoing and
currently inconclusive. A perspective led by constraints proposes that there are
many route maps to a high level of performance. All in all, these arguments suggest that a major focus of research should be on understanding the differences
among top athletes rather than on overemphasizing perceived commonalities,
such as time and amount of practice, that influence the acquisition of expert
skill in movement (Davids, 2000).
Perhaps the most critical challenge facing proponents of the theory of deliberate practice concerns the role of genes in constraining responsiveness to
prolonged training and practice. Regarding possible innate qualities that may
predispose an individual to successful performance in a domain, Ericsson and
colleagues have conceded only one qualityheightas being beyond circumvention through deliberate practice. Further, Ericsson et al. (1993) provided
evidence that several characteristics thought to be innate were in fact trainable.
For example, while the ability to distinguish and name the 64 notes in music,
referred to as perfect pitch, is difficult for adults to acquire, it is relatively easy
for children to attain (Takeuchi & Hulse, 1993). This finding lends support
to the assumption of Ericsson et al. that proper training at appropriate times
of development is crucial to expert performance. Moreover, Ericsson et al.
(1993) suggested that physiological parameters could also be modified through
deliberate practice. Research indicating that endurance athletes develop a cardiovascular adaptation known as athletic heart syndrome (George, Wolfe, &
Burggraf, 1991) that reverts to near normal when training has stopped may
denote the influence of training rather than genetic predisposition.

Genetic and Environmental Constraints on Variability in Sport Performance

In the remainder of this chapter we examine the evidence for genetic constraints on performance variability. Numerous genes that contribute to variability
in performance are being identified in the literature on molecular biology and
exercise and sport physiology (for reviews see Davids, Glazier, Arajo, & Bartlett,
2003; Frederiksen & Christensen, 2003). Although certain general traits have
been linked to heritability (e.g., intelligence) (T.J. Bouchard, 1997), it is widely
accepted that the refinement of these traits into domain-specific abilities (e.g.,
pattern recognition, strategic thinking) occurs through exposure to optimal
preparation in specific environments. There is little evidence to support the
idea that there is a single gene predisposing an athlete to superior performance
in a specific domain (e.g., a gene for hand-eye coordination or a genetic predisposition to play ball games), and the application of this idea has begun to
occur in support services for elite sports performance (Dennis, 2005).

Genetic Constraints on Physical Performance

The search for the genetic basis of many human capacities such as physical
performance has engendered strong arguments in the literature, with some
molecular biologists calling it the biological counterpart to the holy grail
(Kevles & Hood, 1992) and some sport scientists asserting that genetics are
responsible for up to half of the variation in physical performance among
individuals within a population (Hopkins, 2001). Interpretation of the extant
literature is complex because the research on genes and physical performance
expands almost weekly and there is considerable rhetoric among the genuine
conclusions that can be drawn. It can be concluded that there is a great deal
of equivocality in the existing research on the genetic basis of physical performance (Davids et al., 2003).
Nevertheless, it is possible to interpret existing data on interindividual
variability in health and performance based on the interaction of genetic and
environmental constraints. For example, adiposity is considered a constraint
on performance in some sports and physical activities, and although increasing
adiposity (within limits) may not harm performance in certain sports such as
Sumo wrestling or rugby union and league, performance in endurance activities
may suffer considerably from unfavorable levels of adiposity. The interactive
role of genes and environment is emphasized by the growing consensus in the
study of human obesity that the contribution of genetic factors is exacerbated
by different environmental constraints including caloric availability (e.g., Barsh,
Farooqi, & Rahilly, 2000). Genetic propensity toward adiposity has less of a
constraining influence on individuals in environments where caloric availability
is lower, whereas these same individuals would be at greater risk of obesity in
calorie-rich environments. Environments can be categorized as high or low risk,
depending on the prevalence of other significant cultural constraints including the availability of training facilities, work patterns imposed on traditional

119

120

Baker and Davids

mealtimes, and the fall in popularity of physically active pastimes leading to

a greater emphasis on static activities such as playing computer games and
watching TV. Thus, the interaction of genes and environment on the phenotypic
expression of behavior can be best understood by considering individual risk
rather than by considering them as defective behavior (i.e., as in the medical
model).. This is an important point when considering the effects of spending
time practicing in sport. Given interindividual genetic differences, variations
in physical performance are more likely to assert themselves under intensive
practice regimes.
In fact, during the past decade there has been increasing research on the role
of genetics in defining the level of athletic performance attainable by individuals. As we note in the following sections, research has focused on genetic and
environmental contributions to physical (typically endurance) performance
(e.g., Rankinen et al., 2000), although there have been isolated attempts to
evaluate relative contributions to the acquisition of motor skill (e.g., Fox,
Hershberger, & Bouchard, 1996).

Genetic Contribution to Motor Skill Performance

L.R. Williams and Gross (1980) studied the performance of 22 monozygotic
(MZ) and 41 dizygotic (DZ) twins on a stabilometer balance task over 6 d to
examine the genetic contribution to learning and performance. The prediction
was that interindividual variation in performance and learning would be less
in the MZ group as compared to the DZ group. This prediction was supported
by data indicating a greater intrapair resemblance in the MZ group only when
the learning profiles of the twins were compared over time. Intraclass correlations were used to estimate the proportion of the total phenotypic variance in
performance and learning that was accounted for by heritability. Heritability
effects were reported to be low during the earliest stages of learning, but they
became increasingly powerful as practice continued. Furthermore, the proportion of variance in performance accounted for by systematic variation of the
environment due to the manipulation of constraints by coaches and teachers,
was highest during the early stage. Although heritability is made up of genetic
and environmental components, these findings imply that there is potential
for influencing performance and learning by manipulating task constraints
during practice.
Other work has been more ambitious in its aims. A study of performance
in pursuit rotor tracking by Fox and colleagues (1996) examined the performance and learning of MZ (n = 64 pairs) and DZ (n = 32 pairs) twins reared
apart. Performance outcome was scored by the time spent on target over 75
trials and was expressed as a proportion of the perfect score, 20 s. Fox et al.
(1996) observed that the performance of the groups was very similar, with both
substantially improving over the five trial blocks of the first day. Patterns of

Genetic and Environmental Constraints on Variability in Sport Performance

variability for both groups were also similar. Over practice, some participants
improved more than others, which led to increases in variability within groups
by the third day of the practice regime. However, statistical analysis did not
reveal significant differences between the variances of the MZ and DZ twins
over trials. The authors noted that there was greater variability in correlations
with task performance in the DZ group over trials, although this effect may
have been partly due to the smaller number of DZ pairs studied. The slope of
the regression line for the DZ intraclass correlations for the last 2 d was close
to zero, implying that the contribution of environmental factors decreased as
practice continued. Despite the large intergroup differences in the number of
participants, the authors concluded that the consistently larger intraclass correlations for performance in the MZ group as compared to DZ group pointed
to a significant genetic component of performance (see figure 6.4).
The authors proposed that a model combining genetic and environmental
effects best fit the data. The influence of heritability (reflecting both genetic and
environmental factors) was high from the first trial block (proportion of contribution to performance variance = 0.66) to the last trial block (proportion = 0.69).
The fact that the influence of heritability was high for the first of the initial 5
trial blocks (0.66, 0.53, 0.52, 0.55, and 0.52, respectively) might be taken as
evidence that individuals rely on innate capacities for the first few practice trials
of a novel task. Conclusions by the authors of a clear distinction between MZ and
DZ for dependent variables such as percent time on target, rate of improvement
of performance over trial block, and improvement after a time of rest were based
on genetic influence. More work is needed, however, since the authors seem to
confuse performance with skill acquisition. Although skill acquisition is the
phrase used in the title of the paper, only 75 trials were examined and it could
be argued that performances of both groups were measured.
The issue of the potential confounding effects of unequal sample sizes in
the study by Fox et al. (1996) is nontrivial. In complete contrast to the findings obtained by Fox et al. (1996), other work examining differences in the
performance of pursuit tracking between equal numbers of pairs (n = 35) of
MZ and DZ twins proposed that the strength of the genetic constraints on
performance systematically diminished throughout the course of practice, fitting a monotonic trend over trials (Marisi, 1977). Joseph (2001) has outlined
a number of other methodological concerns with studies on twins. Classical
methodology in research with twins compares the correlation or concordance
rates for measurements from same-sex DZT (dizygotic, reared together) and
MZT (monozygotic, reared together) twins. Identical (MZ) twins share 100% of
the same genes while fraternal (DZ) twins share only 50% on average. Greater
similarity in MZT twins is taken as evidence of the powerful influence of
genetic constraints. The assumption is that both types of twins share the same
environment, although it has been argued that data are confounded by MZT
participants having a greater environmental similarity than DZT participants
(Joseph, 2001).

121

122

Baker and Davids

FIGURE 6.4 Data from Fox et al. (1996) on the performance of pursuit rotor tracking in monozygotic (n = 64 pairs) and dizygotic (n = 32 pairs) twins reared apart.
The top graph purports to show a high influence of heritability on skill practice in
the first few practice trial blocks and the maintained influence of heritability over
trials. The bottom graph shows the differences between monozygotic (open squares)
and dizygotic (filled squares) twins in the magnitudes of intraclass correlations for
performance, indicating that there is a significant genetic component of performance.
Reprinted, by permission, from P.W. Fox, S.L. Hershberger, and T.J. Bouchard, 1996, Genetic and environmental contributions
to the acquisition of a motor skill, Nature 384: 356-358.

Genetic and Environmental Constraints on Variability in Sport Performance

Another favorite methodology is to study MZA (monozygotic, reared apart)

twins separated at birth and raised under different socioeconomic and cultural
constraints. Such a comparison is believed to provide an ideal analysis of the
effects of nature and nurture. Genetic inferences from studies on separated
twins are based on the assumption that the shared environments of the twins
were not systematically more similar than those of unrelated and randomly
paired individuals, the so-called unequal environment assumption. According to
Joseph (2001), the problem is that comparisons of separated twins reared in
distinct environments are almost impossible to achieve in reality. He argued
that there are many difficulties in obtaining pure samples that fit the stringent
criteria needed for this type of test of genetic and environmental constraints.
Twin studies adopting this methodology can be contaminated in many different ways including:
The twins are separated only after being raised together a long time
(years).
They are raised by members of the same family.
Their placement families are correlated for many factors to ensure equitable living conditions.
In contrast to the assumption of minimal contact, the twins remain aware
of each other and maintain contact.
They are brought to the attention of researchers because they are perceived
to be very similar and worthy of further study.
Data on both twins are collected by the same researchers rather than by
independent observers, which leaves the data open to bias and expectation.
Often an important error in twin studies is the assumption of different socioeconomic and cultural backgrounds, which is difficult to achieve because of
historical constraints. That is, two people born on the same day and brought
up at the same time and in the same culture, possibly sharing similar class and
ethnic values, may be expected to show a great deal of similarities because of
the so-called cohort effect (a constraint on group affiliations). One important
way in which cohort effects can be ruled out as an explanation for the data
of studies on twins is using closely matched pairs of biologically unrelated
strangers as controls to the MZA participants. One difficulty in interpreting the
data from studies on twins is that information about participant recruitment
is often not reported and there is a lack of case histories to help independent
judgments of the data.
These significant problems have led to the conclusion that there has not been
a clear demonstration that MZA twins are reared in uncorrelated environments
to support the unequal environment assumption. According to Joseph (2001),
significant MZA personality and behavioural correlations can be explained

123

124

Baker and Davids

plausibly on the basis of the various environmental similarities shared by

separated identical twins and by inflated figures resulting from bias and error
in the various studies (p. 24).

Genetic Contribution to Variability in Physical

Performance
Not all research on genetic constraints has adopted the twin studies approach.
The significant interindividual variations observed in response to training of
the cardiovascular system has led many investigators to question the extent
to which genetic diversity may be responsible for the data (e.g., Feitosa et al.,
2002). In the past few years, the role of the ACE1 gene has received considerable attention in the literature on exercise physiology, molecular biology, and
sports medicine. In the next section, we evaluate the evidence for its role as a
genetic constraint on variation in physical endurance.

The ACE Gene

The ACE gene is one of a number associated by research with interindividual
variability in performance in physical endurance (Alvarez et al., 2000; Montgomery et al., 1999; Montgomery & Payne, 2004; Myerson et al., 1999;
Nazarov et al., 2001; Taylor, Mamotte, Fallon, & van Bockxmeer., 1999; A.
G. Williams et al., 2000; A.G. Williams et al., 2004; Woods, Humphries, &
Montgomery, 2000; Woods et al., 2001; Woods et al., 2002). In muscle, the
angiotensin I-converting enzyme degrades vasodilators (i.e., bradykinin and
tachykinin) and stimulates production of the vasoconstrictor angiotensin II
during physical performance (Sonna et al., 2001). To date, three variants of
the human angiotensin I-converting enzyme (ACE) gene have been found.
The presence or absence of a fragment containing 287 base pairs characterizes the I (insertion) or D (deletion) allele, respectively, leading to 3 variants
(II, ID, and DD).
Increasing ACE activity is linked with the D allele, affecting the degradation of
bradykinin and the synthesis of angiotensin II. DD participants show increased
conversion of angiotensin I to angiotensin II, the latter having a vasoconstriction effect. However, angiotensin II seems to stimulate endogenous factors
for the growth of muscle cells, contributing to a hypertrophic response useful
for power development. Degradation of bradykinin results in lower substrate
metabolism and less efficient vasodilation. Therefore, lower ACE activity may
be associated with an increased half-life of bradykinin that alters substrate
metabolism. Increased angiotensin II is associated with the DD genotype and
may facilitate muscle bulk for power sport performance. It is estimated that
1

ACE stands for angiotensin-converting enzyme.

Genetic and Environmental Constraints on Variability in Sport Performance

25% of the population have the II genotype, 50% the ID genotype, and 25%
the DD genotype (Jones, Montgomery, & Woods, 2002).
One approach to research on the ACE gene has been to examine whether a
particular genotype occurs more frequently in specific populations as compared
to controls. If a polymorphism is found to prevail more in a specific population
as compared to matched controls, then either the polymorphism or its locus
on the chromosome may be responsible for different frequencies of appearance. Alternatively, the polymorphism may be in linkage disequilibrium, that
is, closely associated with a different locus on the chromosome that is actually
responsible.
For example, the earliest work with army recruits found that the genotype
II polymorphism of the gene is associated with lower ACE activity in muscle
and an increased response to physical training (Montgomery et al., 1998).
Recruits with the ACE genotype II differed by as much as 1,100% in response to
repetitive upper-arm exercises when compared to peers with the DD genotype.
Individuals with a heterogeneous genotype (DI) were associated with levels of
performance between those of both homozygous genotypes. In sport, a higher
prevalence of the II genotype has been found in elite endurance athletes including mountaineers able to climb to 7000 m without the aid of oxygen, Olympic
endurance runners, and elite rowers (Gayagay et al., 1998; Montgomery et al.,
1998; Myerson et al., 1999).
Interpreting the data from studies on the genetic constraints in physical
performance and in the acquisition of motor skill is rather complex and there is
enormous potential for confusion amidst the rhetoric. Initially, the data favoring
a strong genetic constraint on physical performance seemed compelling. While
most researchers studying genetic variations in human performance agree with
Hopkins (2001) opinion that athletes are born and made, a clear interpretation
of the data on the ACE gene is needed to understand how athletic performance
emerges under interacting constraints.
A good example of the appropriateness of this conclusion in the face of rhetoric
that human physical performance is strongly influenced by genetic factors (e.g.,
Myerson et al., 1999) was provided in a study by C. Bouchard and colleagues.
They attempted to establish the proportion of influence attributable to genetic
and environmental
constraints on familial resemblance for maximal oxygen
.
uptake ( VO2max) during exercise on a cycle ergometer in sedentary individuals
(C. Bouchard et al., 1998). For this purpose, the exercise performance of fathers,
mothers, sons, and daughters was measured in 86 nuclear families. Maximum
heritability including genetic and nongenetic causes for physical performance
accounted for 51% of the total adjusted phenotype variance. Several models
of interacting constraints were tested, and results showed that there was 2.6 to
2.9 times more variance between families than within families.
Unfortunately, the approach taken in this study meant that genetic and familial environmental influences could not be fully quantified separately, although
inferences about their respective contributions to the phenotype variance could

125

126

Baker and Davids

be made by inspection of the pattern of familial correlations (p. 255). As is

well known, however, correlations do not imply causation. The emphasis on
the constraints imposed by the shared familial environment is also important.
While this explanation for heritability of maximal oxygen uptake may be valid,
it does not preclude the influence of wider environmental constraints such as
sociocultural changes in society, including effects of media images, changing
fashions in society, government education programs, and peer pressure.
Despite the fact that the maximal heritabilities reported in this study were
inflated by familial, nongenetic contributions, the effects of the maternal transmission of mitochondrial DNA to the fertilized zygote were seen as optimally
allied to the fathers environmental contribution. The authors argued that the
data revealed that maternal influence, perhaps by mitochondrial inheritance,
accounts for as much as 30% of the familial transmission (p. 257). The authors
conclusion was that based on the present results, we estimate that mitochondrial heritability is in the range of 30-35% (p. 257). This highly speculative
interpretation of the data is based on correlational statistics, a limited range of
environmental constraints considered as affecting model construction, and no
evidence from DNA analysis.
This study exemplifies the complexities involved in understanding genetic
and environmental constraints on physical performance and in order to enhance
understanding of the literature on genetic constraints on behavior, it is worth
reiterating what is already known in this area of work. As most geneticists
studying physical performance understand, genes work in combination to
influence biological function. This understanding refutes the idea of successful athletes being differentiated by the presence of a single gene (for a similar
argument in developmental theory see Johnston & Edwards, 2002). It has also
become clear that genes are not biologically determinate, since even the most
ardent geneticists agree that the transmission of genetic information between
generations is less than perfect (e.g., Jones, 1999). DNA is simply a copy of
information that is read by cellular machinery in the production of proteins
that create the individual part by part. However, somewhere along the line the
view of DNA as an information bearer has been replaced with the fallacy of
DNA as a plan or master molecule (Lewontin, 2000).

Genes and Variability in Movement Systems

Therefore, the presence of genetic material should not be viewed as a blueprint
for success in sport. As Johnston and Edwards (2002) have pointed out, it is a
very long step from polypeptide sequences to behavioura step . . . that covers
much incompletely understood territory (p. 26). An attempt to see genes as
building plans is one of the great artificialities in human conceptualizations of
nature (van Geert, 1994), but it has become a central dogma of how people
think about the process of evolution (Oyama, 2000). Genes simply contain
the information to synthesize proteins with properties that lead to clustering.
Lewontin (2000) criticized biological determinism, the rejection by the medi-

Genetic and Environmental Constraints on Variability in Sport Performance

cal model view of polymorphism and the implicit notion of variability as deviation from a perfect ideal. Genetic diversity is the norm and biological systems
are not determined by DNA. There is no single, standard DNA sequence that we
all share, and estimates are that we differ in DNA sequencing by 0.1% (about
3 million nucleotides), including sequences inherited from parents. It takes
more than DNA to produce a living organism, and those other components
cannot be computed from DNA sequences. According to Lewontin (2000),
a living organism at any moment of its life is the unique consequence of a
developmental history that results from the interaction of and determination
by internal and external forces (p. 147).
A major argument against the conceptualization of genetics as a blueprint
is found in evidence that identical twins are not actually identical. A study of
phenotypically identical twins showed that their fingerprints differ and that the
shape of their brains can differ by as much as 40% (Yates, 1993). Heritability of
a trait is constrained by genetic and environmental factors to some extent, and
research in behavioral genetics is concerned with explanations of hereditary
influences at the level of populations, not individuals.
Nonetheless, the evidence linking the ACE gene and physical performance
continues to accumulate. Although some work on endurance performance
in elite athletes has failed to support the more functional role of the I allele
of the ACE gene (e.g., Taylor et al., 1999), this study was made up of 120
performers chosen from sports with task constraints emphasizing a high level
of aerobic fitness (including 26 hockey players, 25 cyclists, 21 skiers, 15 track
and field athletes, 13 swimmers, 7 rowers, and 5 gymnasts). An alternative
explanation for the data is that such a mixed group of athletes may not have
had the requisite levels of phenotypic homogeneity to lead to valid estimates
of the genetic basis of performance. Moreover, it has become clear that carriers of the D allele have an advantage in training and performance when
task constraints emphasize power over a shorter duration (Myerson et al.,
1999; Nazarov et al., 2001). In fact, the D allele has been related to increased
gains in quadriceps strength following 9 wk of isometric training (Folland
et al., 2000). It is possible that the D allele may confer some performance
and training benefits in task constraints requiring power (perhaps through
its effect on greater angiotensin II and muscle hypertrophy), and similarly
the I allele may have an effect under task constraints requiring endurance.
The implication is that variability at the level of individual genes provides
functionality and adaptability in movement systems that need to perform a
variety of activities in a complex environment. This suggestion also emphasizes
that in experiments on variants of the ACE gene and sport performance, a
clear understanding of differences in task constraints is needed to ensure that
homogenous cohorts of athletes are carefully examined in order to avoid the
loss of genetic association.
Finally, research on the ACE gene is progressing rapidly, and there are some
indications that its role in constraining physical performance may be somewhat

127

128

Baker and Davids

different than originally perceived. For example, there has been some doubt
cast on the relationship between the I allele and responsiveness to endurance
training that was originally proposed in some studies (e.g., Gayagay et al.,
1998; Hagberg et al., 1998; Montgomery et al., 1998). The locus of the ACE
gene has been identified as chromosome
17q23, and genomic scanning for
.
candidate genes for baseline VO2max performance or responsiveness to training failed to confirm evidence of linkage (C. Bouchard et al., 2000). These
findings on a sedentary population were supported by a frequency analysis
that failed to find a relationship between the accumulations of alleles I and II
and endurance performance in 192 elite athletes (skiers, runners, and cyclists)
and 189 controls (Rankinen et al., 2000). Interestingly, the highest frequencies reported for both the elite athletes and controls were for the ID genotype
(0.46 and 0.47, respectively). Nevertheless, future research needs to ascertain
(a) whether or not the effect of the I allele of the ACE gene on endurance performance is mediated via peripheral muscle effects and changes in efficiency,
and (b) whether or not the effect of the D allele on performance in power
tasks is mediated via increased angiotensin II acting as a local hypertrophic
factor in muscle.
To summarize, the main difficulty with current research on the ACE gene
is that investigators seem to have conducted research on samples with mixed
phenotypes leading to equivocality of findings. Sometimes, the label given to
specific populations has not been accurate (e.g., elite versus subelite athletes)
(Jones et al., 2002). The strongest associations between II and DD polymorphisms and endurance and power performance, respectively, have been found
in homogenous cohorts of elite athletes of specific sport disciplines. The conclusion by Jones et al. (2002) is that the ACE I/D polymorphism should not
be considered a gene for human performance, but a marker for modulation
such that one would expect an excess of the I allele in the truly elite endurance athlete, with a concordant excess of the D allele represented in the more
power-oriented events. Therefore, the study of mixed cohorts is unlikely to
prove fruitful (p. 187).
One problem with this explanation for equivocality by Jones et al. (2002)
is that it is post hoc. That is, there is a question mark over the predictive power
of using the ACE gene polymorphism to explain performance in endurance
and power sports. It seems that the linkage is clear only with pure samples
of elite athletes, and where no effects are found it might be possible to argue
that the samples were not pure. The lack of clarity in the literature was confirmed by Jones et al. (2002), who stated that the ACE genotype has never
been associated with endurance performance in the untrained state. Any effect
appears to require a period of gene-environment interaction. A high level of
aerobic fitness is an essential, but not sole, requirement for elite endurance
(p. 188). A final point is that there is a high level of individual variation in
the data on the ACE gene and endurance and power performance. Jones et
al. (2002) argued that there will always be elite endurance athletes who are

Genetic and Environmental Constraints on Variability in Sport Performance

of the ACE DD genotype, and many champions in anaerobic sports of the II

genotype. Whatever the data may conclude, elite athletes are still made and
not born, though perhaps some may be made elite in one discipline more
easily than others (p. 189).

Concluding Remarks: A Case for Dynamic

Systems Theory
We have evaluated the strengths and weaknesses of theoretical ideas and empirical research for theories of learning and performance that posit major effects for
environmental and genetic constraints. It was concluded that neither approach,
each emphasizing the unitary role of one category of constraints, provided
enough explanatory power to account for data on variability in performance,
suggesting it may be premature to include genetic testing as part of athletic
screening programs (e.g., Dennis, 2005). It was noted that the implicit basis
of the perspective of deliberate practice is the adage all individuals are created
equal. The analysis of the literature on genetic constraints on variability in
performance does not support this conclusion, but this analysis should not be
taken to imply that performance is biologically determined. Rather, the effects
of interacting constraints on health and performance have been noted, since
despite variations in genetic structure, the maximal heritability of particular
traits includes strong environmental components.
A theoretical perspective based on dynamic systems, in which interacting
constraints explain variability in behavior, may provide an adequate overarching framework for interpreting data (Davids et al., 2003; Davids, et al., 2004).
Genetic diversity may be responsible for a small part of the differences in
training or performance response in individuals and performance benefits may
be observed only when there is a favorable interaction with important environmental constraints. The effects of the available environment on phenotypic
expression were noted in research on causes of obesity in human health. The
implication of these findings and of data from studies on endurance performance
is that elite athletes of a less favorable genotypic disposition can succeed with
the appropriate training environment. However, it can be concluded that performers with a more favorable genotype who appropriately interact with their
training environments are more likely to receive a greater response to training.
The current data on genetic constraints in the acquisition of motor skills are
unclear due to various methodological weaknesses and conflicting findings,
and more work is needed to identify genetic mechanisms underlying variations
in performance. Moreover, the emphasis on constraints placed by dynamical
systems theory implies new ways of looking at the whole nature versus nurture
argument. This theoretical perspective provides an overarching framework that
encompasses an extensive variety of organismic and environmental constraints
on human behavior.

129