Intelligence and chess
Fernand Gobet, & Guillermo Campitelli
Psychological research into expert behaviour in board games is currently split
into two, almost independent, strands. The first direction of research, initiated by
De Groot’s (1946) and Simon and Chase’s (1973) seminal research on chess, is
interested in the cognitive processes underlying experts’ remarkable feats and in
the learning mechanisms that enable novices to move to high levels of performance.
The second line of research, exemplified by the early work of Alfred Binet on
chess (Binet, 1894), is more interested in the innate characteristics that are putatively
necessary for high ability.
By and large, our own research has focused upon the first of these strands, and
has led to the development of several computational models of chess expertise (De
Groot & Gobet, 1996; Gobet, 1993, 1997, 1998; Gobet & Simon, 1996, 2000).
These models, built in the tradition of Simon and Chase, suggest the following
requirements for becoming an expert: acquiring a large number of perceptual pat-
terns (more than 100,000 in the case of chess), linking them to information such as
strategic themes, tactical motives, potential moves, and using them during look-
ahead search. As a consequence, these models emphasise the role of practice and
study. However, these models, which contain various parameters such as the capacity
of short-term memory or the time to learn a Ôchunk’ of information, also suggest
that there are innate individual differences between players.
The goal of this paper is to critically evaluate the evidence supporting the
hypothesis that innate talent (for example a higher level of intelligence) may account
for aspects of chess skill. We start by giving some standard definitions of intelli-
gence, and then present empirical data on chessplayers’ intelligence, both with
children and adults. In the second part of this paper, we discuss Geschwind and
Galaburda’s (1985) influential theory of the neurobiology of talent in music,
mathematics, and visual arts, and review evidence of its applicability for explaining
chess expertise. For that purpose, we will present empirical data based on brain
lesions, brain-imaging studies, and handedness. In the conclusion, we will attempt
to reconcile the strand of research emphasising practice and the role of the
environment with that emphasising the role of innate talent.
Expertise in chess
Mainstream cognitive research into chess expertise (there is not much research
into other board games) was started by De Groot’s (1946) work on the decision-
making processes of chessplayers. De Groot identified the critical role of percep-
tion, which allows rapid access to information stored in long-term memory. Buil-
ding on this research, Simon and Chase (1973) proposed a detailed theory of the
cognitive mechanisms involved in chess playing, and, in particular, specified the
learning mechanisms allowing acquisition of perceptual knowledge. Several as-
pects of this theory and of its revisions have been implemented as computational
models, which closely replicate empirical data about eye movements, memory per-
formance, and search behaviour (De Groot & Gobet, 1996; Gobet, 1993, 1997,
1998; Gobet & Simon, 1996, 2000; Simon & Barenfeld, 1969; Simon & Gilmartin,
1973). Pattern recognition plays an important role in this line of research: through
years of practice and study, masters have learnt several hundred thousands of
perceptual patterns, which, once recognised in a particular position, give rapid access
to information such as potential moves or move sequences, tactics, strategies, and
so on. Simon and his colleagues proposed that pattern recognition explains a number
of important phenomena, such as highly selective search (even chess grandmasters
rarely search through more than one hundred moves before selecting a move),
automatic and “intuitive” discovery of good moves, and extraordinary memory for
game-like chess positions. Simon and Chase (1973) suggested that at least ten
years of practice and study were necessary to acquire the minimum knowledge
required to become a grandmaster.
As already mentioned, the computer models developed in the Simon and Chase
tradition leave ample room for individual differences: these models include various
parameters, such as the capacity of short-term memory, the time to learn new infor-
mation, or the time to store information into short-term memory. While these
parameters have been set using empirical data aggregated across subjects, it is
reasonable to assume that these values may vary between individuals (however, for
a strong dissent with this assumption, see Ericsson, Krampe & Tesch-Römer, 1993).
Indeed, anecdotal evidence would suggest the presence of such individual
differences. For example, while many chess players train very hard to become the
next Bobby Fischer or Garry Kasparov, only a handful of them reach grandmaster
level. And among the minority of those who are successful, there are obvious
differences in the time needed. Some players require more than 20 years (e.g., Pal
Benko; Paul Van den Sterren), while others need only 10 years (e.g., Bobby Fis-
cher, Judith Polgar) or even less (Ruslan Ponomariov, who became grandmaster in
1997, needed only about 7 years).
Intelligence and Chess 107
In addition, there are important individual differences in the style of play:
some players are aggressive, others defensive; some prefer tactical complications,
others transparent strategic planning. Finally, one can look at extra-chess activities
for evidence of individual differences. In his 1946 book, De Groot found that there
were important differences in training and background in the sample of 55
grandmasters he studied. In particular, he found that 13 of his grandmasters had a
training in science or mathematics. Interestingly, such differences in background
have tended to fade away in recent years: nowadays, with the stringent training
requirements of competitive chess, most players are professional, with no university
Where do these differences come from? Several psychological explanations,
paralleling the strands of research mentioned above, have been advanced. Infor-
mation-processing research tends to emphasise the role of the environment (presence
of coach or playing opportunities, coaching techniques, etc.). The extreme position
in this strand has been taken by Ericsson (e.g., Ericsson et al., 1993) in his theory of
deliberate practice, which denies the role of innate differences, except for motiva-
tion and the ability to sustain long-term practice.
On the other side of the divide, several theories have emphasised the role of
innate talent, proposing that individual differences are due to variations either in
general intelligence or in specific aspects of intelligence. Variations in general in-
telligence may in turn be due to differences in speed of information processing or
in the quality of signal transmission in the nervous system. Specific differences
(e.g., in visuo-spatial ability) may be due to differences in development of definite
regions of the brain, such as right-hemisphere. Below, we will consider Geschwind
and Galaburda’s (1985) theory, which relates innate talent to differences in brain
The intelligence of chess players
Definitions and measures of intelligence
While central in several subfields of psychology, the concept of intelligence has
turned out to be rather elusive and difficult to define. Indeed, it has generated its
share of controversies for over one century. Two main schools have dominated this
field. The first proposes that several types of intelligence co-exist and the second
considers intelligence as a unitary concept. A good example of the first approach is
offered by Gardner’s (1983) theory of multiple intelligences, which identifies seven
types of intelligence, such as spatial, logico-mathematical, and musical intelligence.
How the second approach construes intelligence may be illustrated by some
definitions, taken from Mackintosh’s text (1998).
Intelligence is seen as:
-“general mental efficiency” (Burt, 1949)
-“the aggregate or global capacity of the individual to act purposefully,
to think rationally, and to deal effectively with his environment”
-“a general reasoning capacity useful in problem-solving tasks of all
kinds” (Kline, 1991)
Alfred Binet, who incidentally carried out the first study on the mental abilities
of chess masters in 1894, was the first psychologist to develop an intelligence test
(Binet, 1903).1 His influence is still visible in some of the tasks used in intelligence
tests, and even in their name: intelligence tests are often called Intelligence Quo-
tient (IQ) tests, a remnant of Binet’s work where the intelligence score was the
quotient of mental age to physical age.
A variety of tests have been designed to measure intelligence, typically adopting
a compromise between unitary intelligence and multiple intelligences, and providing
measures both for various components of intelligence and for general intelligence.
For example, the popular Wechsler test (Wechsler, 1944) has a scale for verbal IQ
and a scale for performance IQ (i.e., non-verbal IQ); in turn, several tasks of per-
formance IQ directly tap visuo-spatial IQ. Finally, a composite score measures
general IQ. Measures of verbal IQ include subtests tapping general knowledge,
knowledge of vocabulary, and memory for numbers. Measures of performance IQ
contain subtests such as arranging a number of pictures so that they tell a coherent
story, matching as many digits as possible with the appropriate symbols, and so on.
Performance IQ also contains measures of visuo-spatial IQ, such as completing
pictures which have one missing part, forming certain designs with nine coloured
blocks, and simple jigsaw puzzles. It is worth mentioning that most of the perfor-
mance IQ tests either have a time limit or offer time bonuses for rapid completion.
Children’s IQ and chess
Two studies have investigated the potential relationship between chess skill and
intelligence. Frydman and Lynn (1992) submitted 33 young Belgian chess players
(mean age = 11 years) to the Wechsler test. They found that the average general IQ
of their sample (121) was higher than the population mean (100); most of the effect
was accounted for by the performance IQ (mean = 129), which was clearly higher
than the verbal IQ (mean = 109). Finally, there was some evidence that better players
had higher performance IQ scores than the weakest players (top third = 131 vs.
bottom third = 124).
The second study was carried out by Frank and D’Hondt (1979), who randomly
allocated 90 children from Zaire (around 14 years old) either to a chess playing
class or a control class, for one year. Four psychometric tests, totalling 18 subtests,
Intelligence and Chess 109
were given to the children both before and after the intervention. The design allowed
Frank and D’Hondt to test both whether some aptitudes predict chess skill after one
year, and whether chess training might improve some of these aptitudes.
There was evidence that some abilities do predict chess skill. Five sub-tests
were found to reliably predict chess skill at one year: Ôspatial aptitude’, Ônumeric
ability’ (2 sub-tests), Ôadministrative sense’, and Ôoffice work’. There was also
evidence that chess skill may improve aspects of intelligence. The treatment group
did better than the control group in the post-test on the sub-tests Ônumerical ability’
and Ôverbal ability’.
Adults’ IQ and chess
In 1927, Djakow, Petrowski, and Rudik (1927) studied eight of the best
grandmasters of the time. They did not find differences with a control sample on
general intelligence or visuo-spatial memory, with the exception of memory tasks
where the material to be recalled was closely related to chess. More recently, Doll
and Mayr (1987) subjected a group of chess masters (N = 27, age = 25.7 years) and
a control group of non-chessplayers (N = 88, age = 24.8 years) to the “Berlin
Structural Model of Intelligence” test, a well validated IQ test. The masters did
better in the measure of general intelligence, as well as in tasks related to
ÔInformation-processing capacity for complex information’, ÔWorking speed’,
and ÔNumerical thinking’. Surprisingly, given the fact that chess is a highly visuo-
spatial game, the masters did not do better in the visuo-spatial task.
Further research into visuo-spatial abilities in chess gave mixed results.
Interference tasks with chess problem solving and memory (Saariluoma, 1992)
showed that a visuo-spatial concurrent task has strong effect, but that a verbal con-
current task has no effect. However, Waters, Gobet & Leyden (2000) found no
evidence for higher visuo-spatial with chess players in a memory task using non-
In summary, research into intelligence and chess has uncovered good evidence
for higher general and performance IQ for chess players. While this could be
explained by the talent hypothesis, it could also be explained by the fact that
chessplayers receive a lot of practice in thinking under time pressure, as all
competitive chess is played using a clock in order to limit thinking time. Thus,
chessplayers may have acquired generalisable skills about how to handle time
constraints. However, no superiority was found with chess playing adults on visuo-
spatial tasks, while there was some evidence of superiority with chess playing
children on visuo-spatial tasks.
The neurobiology of chess talent
Geschwind and Galaburda’s theory (1985)
Geschwind and Galaburda (1985) developed an influential theory of the
neuroanatomical substrate of talent. Their aim was to explain a complex pattern of
results linking, among other things, phenomena such as talent in visuo-spatial
domains (e.g., music or mathematics), brain lateralisation, dyslexia, proneness to
allergies, and handedness. The theory is rather complicated, and, given limits in
space, we will limit ourselves to its main components. It is known that the right
hemisphere of the brain normally underpins visuo-spatial abilities (e.g., Kosslyn &
Koenig, 1992). Geschwind and Galaburda reasoned that better development (e.g.,
pattern of cortical connections) of the right hemisphere should lead to better per-
formance in visuo-spatial tasks. The key step is to propose that great exposure or
high sensitivity to intrauterine testosterone in the developing male foetus leads to a
less developed left hemisphere than usual, and, as a compensation, to a more
developed right hemisphere. Hence, males should be more represented than females
in visuo-spatial domains such as mathematics, music, and chess, and left-handers
should be more represented in these fields than in the general population. We now
consider the available evidence for testing Geschwind and Galaburda’s theory in
Effects of brain lesions on chess skill
A direct consequence of the theory is that lesions to the right hemisphere of the
brain should affect chess skill more than lesions to the left hemisphere. Cranberg
and Albert (1988) collected data about eight players with chess activity before
brain damage. They found that the ability to play chess is preserved in patients
even with large left-hemisphere lesions, and that small right-hemisphere lesions
did not affect chess skill either. However, because they did not present evidence
for the effect of large right-hemisphere lesions, their study is inconclusive.
Images of the chess brain
Another consequence of the theory is that chess playing should engage the right
hemisphere more than the left. Nichelli et al. (1994), using positron emission
tomography (PET), studied brain activation in 10 right-handed males, who had
been playing chess for more than 4 years. They used simple tasks, such as Black/
White discrimination, spatial discrimination, rule retrieval, and checkmate
judgement. They found that these tasks called for the activity of a network of several
interrelated, but functionally distinct, cerebral areas, but did not uncover any
evidence for a predominant role of the right hemisphere.
Onofrj et al. (1995) used single photon emission computerised tomography
(SPECT) to study brain activation in a more complex task: trying to solve a chess
problem mentally. They found non-dominant dorsal-prefrontal activation and also
Intelligence and Chess 111
a lower non-dominant activation on the middle temporal cortex. These results are
in line with what is known about these two areas from the study of different tasks:
the dorsal prefrontal cortex is typically activated in problem solving activities
involving planning, and right mid-temporal lobe activation is typically observed
during memory retrieval of non-verbal information. As predicted by Geschwind
and Galaburda’s theory, the four right-handers presented activation in the right
hemisphere. However, contrary to the prediction, the left-handed subject presented
similar activation in the left hemisphere. An important limit of this study is that
Onofrj et al. used only one position, and that this position (Lasker-Bauer, Amster-
dam, 1889) is likely to have been known to the subjects.
Handedness and chess
Another prediction of Geschwind and Galaburda’s theory (1987) is that the
percentage of non-right-handers should be higher in the chess playing population
than in the general population. Cranberg and Albert (1988) sent a questionnaire to
about 400 US chessplayers figuring in the US Chess Federation ranking list. They
targeted the 200 best US players (Elo rating > 2250), and the 200 weakest amateurs
(Elo rating < 1275). The questionnaire asked players to identify their handedness,
with four options: right-handed, left-handed, ambidextrous, or left-handed as a child
and later switched to right-handed. For the 266 players who answered, they found
that 18% were non-right-handers. This percentage significantly differs from the
percentage in the general population, which has been estimated to lie between 10%
and 13.5%. However, no differences were found between the group of strong players
and the group of weak players, although they were separated by more than five
A weakness of Cranberg and Albert’s (1988) study, however, is that they used a
rather informal method to measure handedness. In a replication, Gobet and
Campitelli (2001) used a well validated questionnaire, the Edinburgh Handedness
Inventory (Oldfield, 1971), with 101 male players in Buenos Aires (from 1490 to
2473 Elo). The results were quite similar to what was found in the previous study:
17.9% of the chessplayers were non-right-handers, and there was no skill difference
between strong and weak players.
Another source of evidence for the biological substrate of talent, which is perhaps
related to the question of handedness, comes from the season of birth. Gobet and
Campitelli (2001) found that the month of birth may affect chess skill: in the Inter-
national rating list of January 2001, grandmasters from the northern hemisphere
were more likely to be born in the first half of the year than in the second half
(57.2% vs 42.8%). This result is also valid with other categories of players figuring
in the international list.
The goal of this article was to review the available data about the possible links
between chess talent and intelligence, and between chess talent and biological
(including innate) mechanisms. The literature on intelligence indicates that chess
players’ IQ is higher than that of the general population, but, surprisingly, does not
offer any evidence that adult chess players have better visuo-spatial skills (some
evidence of higher visuo-spatial skills was found with children). Waters, Gobet
and Leyden (2000) suggest that, while visuo-spatial skills may be important in the
early development of chess skill, other skills, such as motivation, become impor-
tant over time. Note also that the direction of causality is unclear, and each of the
three possible scenarios below is in part supported empirically: (1) chess improves
intelligence; (2) more intelligent individuals play better chess; or, (3) a third varia-
ble (e.g., motivation, ability to think under time pressure) mediates intelligence
and chess skill.
Some predictions of Geschwind and Galaburda’s (1985) theory of talent were
supported by the empirical data. The brain-lesion studies confirmed the prediction
of the relative minor role of the left hemisphere in chess playing, but did not offer
strong evidence for the predominant role of the right hemisphere. The brain-imaging
studies offer some (weak) evidence for the role of the right hemisphere. Finally,
two studies found that the incidence of non-right-handers is higher in the chess
population than in the population at large.
Taken together, these results suggest that there exists biological determinants of
expertise in chess. As mentioned above, there also exists massive evidence for the
role of practice/study as well as for the importance of the environment. The safer
tentative conclusion is that both sources of variability are important. For example,
the famous chess school of Botvinnik, while offering a high standard of coaching
for all its pupils, produced players with widely different skill levels.
The correct question for further research is not to quantify the respective role of
Ônature’ and Ônurture’, as has often been done in the past in research into talent
and intelligence, but to explain how these two components interact dynamically as
a function of time. Given the potential complexity and the dynamic character of
these interactions, we believe that the best way forward is to build mathematical
models or computational systems. Once constructed, such models or systems could
be applied to test the development of talent in other board games as well.
Intelligence and Chess 113
Binet, A. (1894). Psychologie des grands calculateurs et joueurs d’échecs. Paris:
Binet, A. (1903). L’étude expérimentale de l’intelligence. Paris: Schleicher.
Cranberg, L., & Albert, M. L. (1988). The chess mind. In L. K. Obler & D. Fein
(Eds.), The exceptional brain. Neuropsychology of talent and special abilities.
New York: Guilford Press.
De Groot, A. D. (1946). Het denken van den schaker. Amsterdam: Noord
De Groot, A. D., & Gobet, F. (1996). Perception and memory in chess. Heuristics
of the professional eye. Assen: Van Gorcum.
Djakow, I. N., Petrowski, N. W., & Rudik, P. A. (1927). Psychologie des
Schachspiels. Berlin: de Gruyter.
Doll, J., & Mayer, U. (1987). Intelligenz und Schachleistung—Eine Untersuchung
an Schachexperten. Psychologische Beiträge, 29, 270-289.
Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The role of deliberate
practice in the acquisition of expert performance. Psychological Review, 100,
Franck, A., & D’Hondt, W. (1979). Aptitudes et apprentissage du jeu d’échecs au
Zaire. Psychopathologie Africaine, XV, 81-89.
Frydman, M., & Lynn, R. (1992). The general intelligence and spatial abilities of
gifted young Belgian chess players. British Journal of Psychology, 83, 233-235.
Gardner, H. (1983). Frames of mind: The theory of multiple intelligences. New
York: Basic Books.
Geschwind, N., & Galaburda, A. M. (1985). Cerebral lateralization: Biological
mechanisms, associations and pathology: 1. A hypothesis and a program for
research. Archives of Neurology, 42, 428-459.
Gobet, F. (1993). Les mémoires d’un joueur d’échecs. Fribourg: Editions universi-
Gobet, F. (1997). A pattern-recognition theory of search in expert problem solving.
Thinking and Reasoning, 3, 291-313.
Gobet, F. (1998). Expert memory: A comparison of four theories. Cognition, 66,
Gobet, F., & Campitelli, G. (2001). Two markers of chess skill: Month of birth and
Gobet, F., & Simon, H. A. (1996). Templates in chess memory: A mechanism for
recalling several boards. Cognitive Psychology, 31, 1-40.
Gobet, F., & Simon, H. A. (2000). Five seconds or sixty? Presentation time in
expert memory. Cognitive Science, 24, 651-682.
Kline, P. (1911). Intelligence. The psychometric view. London: Routledge.
Kosslyn, S. M., & Koenig, O. (1992). Wet mind. New York: The Free Press.
Mackintosh, N. J. (1998).
IQ and human intelligence. Oxford: Oxford University Press.
Nichelli, P., Grafman, J., Pietrini, P., Alway, D., & al. (1994). Brain activity in
chess playing. Nature, 369, 191.
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh
Inventory. Neuropsychologia, 9, 97-113.
Onofrj, M., Curatola, L., Valentini, G. L., Antonelli, M., Thomas, A., & Fulgente,
T. (1995). Non-dominant dorsal-prefrontal activation during chess problem so-
lution evidenced by single photon emission computarized tomography (SPECT).
Neuroscience letters, 198, 169-172.
Saariluoma, P. (1992). Visuospatial and articulatory interference in chess players’
information intake. Applied Cognitive Psychology, 6, 77-89.
Simon, H. A., & Barenfeld, M. (1969). Information processing analysis of perceptual
processes in problem solving. Psychological Review, 7, 473-483.
Simon, H. A., & Chase, W. G. (1973). Skill in chess. American Scientist, 61, 393-
Simon, H. A., & Gilmartin, K. J. (1973). A simulation of memory for chess posi-
tions. Cognitive Psychology, 5, 29-46.
Waters, A. J., Gobet, F., & Leyden, G. (2000 Submitted for publication). Visuo-
spatial abilities in chess players.
Wechsler, D. (1944). Measurement of adult intelligence (3rd edition). Baltimore:
Williams and Wilkins.
1Binet did not use his test with chessplayers, however.