This paper was published in Autism, 1997, 1, 153-163.
Is there a link between engineering and autism?
Simon Baron-Cohen, Sally Wheelwright, Carol Stott, Patrick Bolton, and Ian Goodyer
Departments of Experimental Psychology and Psychiatry, University of Cambridge
Downing St, Cambridge, CB2 3EB, UK.
Acknowledgements: The first author was supported by the MRC, the Wellcome Trust, the McDonnell Pew
Foundation, and the Gatsby Foundation during this work. We are grateful to Charlotte Russell and Shelagh
Eggo for data coding, and the National Autistic Society, the Tourette Syndrome Association, and Downs
Heart, for their assistance with data collection. We thank Donald Dalton for invaluable guidance in the
field of engineering, and Steve Pinker for stimulating discussion.
Autism is a severe childhood neuropsychiatric condition with a substantial genetic
component. At the cognitive level children with autism are impaired in the development
of their “folk psychology”, whilst they are normal or even superior in the development of
their “folk physics”. We predicted that if their parent shared this cognitive phenotype,
then they should be over-represented in engineering as an occupation. This prediction
was confirmed. Both fathers and grandfathers of children with autism were found more
than twice as often in the field of engineering, compared to fathers and grandfathers of
other children. This link between autism and engineering may throw light not only on
autism itself, but ultimately on the genetic basis of two essential human abilities: “folk
psychology” and “folk physics”.
Autism severely disrupts the normal development of social relationships, communication,
and imagination (APA, 1994). Evidence that it results from neuropathology is plentiful
(Bauman & Kemper, 1994), though the necessary and sufficient aspects of this
neuropathology are not yet known. It occurs at a rate of about 1 per 1000 (Baron-Cohen
et al., 1996; Gillberg, Steffenberg & Schaumann, 1991). From family and twin studies it
appears to have a genetic basis (Bailey et al., 1995; Bolton & Rutter, 1990; Folstein &
Rutter, 1977; Folstein & Rutter, 1988), though the molecular characteristics are not yet
known. The genetic theory of autism however leads to the novel suggestion that autism
may not strike at random, but rather that some types of parents may have an increased
risk of having a child with autism. We report the first large-scale study which tests this
idea in relation to occupations of parents. We predicted that engineers might be over-
represented among the parents and grandparents of children with autism (or the related
condition of Asperger’s Syndrome [AS]). This prediction derives from a theory of
Domain Specificity theory suggests there may be at least 4 universal, “core domains of
cognition” (Carey, 1985; Gelman & Hirschfield, 1994; Pinker, in press; Wellman &
Gelman, in press). These core domains are folk biology (our universal ability to
taxonomize the natural world); folk physics (our universal ability to understand physical
objects in terms of their causal/mechanical properties); folk psychology (our universal
ability to understand the behaviour of other people in terms of their intentional states);
and folk mathematics (our universal ability to count and estimate the probability of
These core domains of cognition appear to be innate in their ‘initial state’, in that they
develop in the majority of humans, irrespective of culture. They constitute a “folk
science” because they are used in an explanatory way by humans. They are considered
‘domain specific’ in that they appear to develop relatively independently of one another,
such that dissociations in the rate of development can be found across individuals.
These 4 domains of cognition were probably of considerable adaptive importance during
the evolution of the brain. That is, possession of each cognitive domain would have
increased the fitness of the individual in different ways. Thus, folk psychology allows for
rapid interpretation and prediction of the actions of other animals, and for social
manipulation. Folk physics allows for tool use in an open-ended way. Folk biology
allows for rapid categorization of individual plants (eg. as edible or inedible), or
categorization of individual animals (eg. as predators or prey). Finally, folk mathematics
allows for estimation of number and probability, essential in planning, for example.
Autism: impaired folk psychology with superior folk physics?
Folk psychology and physics are of special interest in that they involve causal reasoning.
Broadly speaking, folk psychology involves understanding psychological causality: that
people’s actions are caused by their intentional states (their beliefs, desires, knowledge,
intentions, etc.,). Folk physics involves understanding physical causality: that objects
behave in ways that are predictable from a knowledge of physical/mechanical forces.
Here we consider autism in terms of these two types of causal understanding.
Children with autism are known to have major impairments in the development of folk
psychological understanding (Baron-Cohen, 1995; Baron-Cohen, Leslie & Frith, 1985;
Baron-Cohen, Tager-Flusberg & Cohen, 1993) whilst having relatively normal or even
superior development in their understanding of folk physics (Baron-Cohen, in press;
Baron-Cohen, Leslie & Frith, 1986; Frith, 1989; Jolliffe & Baron-Cohen, in press). If
this is a good characterization of their cognitive phenotype, then their parent who carries
the genes for autism might share this cognitive phenotype, to milder degrees (Baron-
Cohen & Hammer, in press a). The prediction therefore is that one would expect parents
of children with autism to pursue occupations in which a ‘talent’ for folk physics is
essential, whilst a talent for folk psychology is not. Engineering is the paradigm case of
such an occupation. This is because it primarily involves a good understanding of
objects rather than people, and is not such a low-frequency occupation as theoretical
physics, for example.1
Method and Participants
1 A different prediction might have been that parents would be over-represented in occupations involving
mathematics and/or computing since these do not necessarily require a talent for folk psychology, but these
do require a talent for folk mathematics or folk physics. However, mathematics is a low frequency
Questionnaires were sent to 1000 parents of children with autism or AS, via the National
Autistic Society (UK) membership list. Parents were asked to list the occupation of the
child’s mother and father, and those of the child’s 4 grandparents. 919 replies were
Similar information was also collected from 4 control groups: (a) parents of children with
Tourette Syndrome (TS), via the Tourette Syndrome Association (TSA) in the UK (n =
40 couples). This served as a control group, to test if patterns of occupations of parents of
children with autism or AS were a function of the sorts of people who become members
of a national charity focusing on a childhood psychiatric disorder. Given that TS can be
associated with autism, only “pure” cases of TS were included. (b) Parents with a child
with Downs Syndrome suffering from cardiac disease, via the charity, Downs Heart, in
the UK (n = 464 couples). Again, this controlled for any sampling bias associated with
being a member of a medical charity. (c) Parents of
children whose language was delayed (n = 98 couples), and (d) parents of children whose
language was not delayed (n =125 couples). These latter two groups provided
data from a random sample, since these 2 groups were derived from community samples.
Information about grandparents was only collected for the autism and TS groups. The
resulting 7,068 occupations were coded blind by 2 independent judges into the 18
occupation, whilst these days computing is part of almost every occupation. Hence our prediction
mutually exclusive occupational categories shown in Tables 1-4, and as defined in the
Legend to the Tables. Inter-rater agreement was 99%.
insert Tables 1-4 here
Results strongly supported the prediction. Fathers of children with autism or AS were
found significantly more often in engineering than fathers in any of the 4 control groups
(Chi Square = p < 0.001). Indeed, fathers of children with autism or AS were found more
than twice as often in engineering, compared to fathers in the control groups. This was
also true of grandfathers of children with autism, compared to grandfathers of children
with TS, suggesting that such effects operate across at least 2 generations in families
where there is a child with autism. The percentage of children with autism or AS who had
a father or grandfather who was an engineer was 28.4%, whereas the percentage of
children with TS who had a father or grandfather who was an engineer was only 15%.
Again, this is a highly significant difference (Chi Square, p < 0.001). Furthermore,
among the fathers of children with autism, the ratio of those working in engineering to
those working in social fields was 6:1, whereas in the 2 charity-based control groups, this
ratio was less than 3:1. This too is highly significant (Chi Square, p < 0.001). There were
no differences in the rate of engineers among fathers or grandfathers of children with
autism versus AS.
This study clearly demonstrates that autism (or AS) does not strike randomly, and
suggests that the cognitive phenotype of fathers of children with autism may be broadly
characterized in terms of their folk physics being superior to their folk psychology. The
finding of an excess of engineers among the fathers and grandfathers of children with
autism is not explained by the social class of the sample, since there were no differences
of this magnitude found in other professions or occupations. Nor is it explained by the
fact that such parents are members of a charity, since it was not found among the fathers
or grandfathers of children with a different disorder (Tourette Syndrome), or among the
fathers of children with Down’s Syndrome. Finally, it appears to be specific to autism
and AS in that it was not seen among fathers of a related childhood condition, language
delay. In all of these control groups, the percentage of fathers in engineering was around
5%, which reflects national levels (Office of Population and Census Survey, 1990).
Mothers and grandmothers of children with autism were not different to female controls.
Note that selection biases appear to be present in the charity-based samples (more doctors
among the fathers of children with TS, reflecting a professional sample, and fewer
absentee fathers in the NAS and TS samples, reflecting the kinds of parents who join
charities); but that neither of these biases can easily explain the link between engineering
and autism across two generations.
There thus seems to be a small but statistically significant link between autism and
engineering. We wish to stress however that the majority of engineers have no
connection with autism, and the majority of parents of autism have no connection with
engineering. Nevertheless, this link between the two phenomena merits further research.
The results of this study fit predictions from Domain Specificity theory, as applied to
autism. The current results might also help explain why a condition like autism persists in
the gene pool: the very same genes that lead an individual to have a child with autism
can lead to superior functioning in the domain of folk physics. Engineering and related
folk physics skills have transformed the way in which our species live, without question
for the better. Indeed, without such skills, homo sapiens would still be pre-industrial.
There is considerable interest in identifying the genetic basis of these two most
fundamental human abilities: folk psychology (also known as mind-reading), and folk
physics (or open-ended tool use), since these two abilities are thought to have played a
major role in primate evolution (Mithen, 1997; Whiten, 1991). The study of the genetics
of autism may throw light not only on the condition itself but also on the molecular basis
of these important human abilities.