ArticlePDF Available

Influence of Experts' Domain-specific Knowledge on Risk Taking in Adversarial Situations

Authors:

Abstract and Figures

Experts play a considerable role in society, as they have to evaluate the risk of policies in many fields of social life and in adversarial situations (e.g., the military). Yet, the influence of expertise on risk taking in adversarial situations has received little attention. An examination of the strategies used by chess players ranging from amateurs to masters in competitive games (n = 73,341) revealed an unexpected pattern of results. First, the majority of players favored the riskier strategy. This result is in line with the literature on economic decision making that indicates a tendency to take risks in situations where the outcome can be either positive or negative. More surprising is our second finding: As skill increased, the majority of players still adopted a risk seeking attitude but the proportion of players taking a more conservative approach increased. This result would tend to indicate that experts making decisions with impact on their own life become increasingly risk averse. Overall, our findings indicate that knowledge moderates but does not eliminate risk taking behavior. They also highlight that risk taking in adversarial situations might result from a complex set of factors. Further research should establish which psychological processes drive players to adopt a risk taking or conservative strategy in their games.
Content may be subject to copyright.
https://www.journalofexpertise.org 88
Journal of Expertise / June 2020 / vol. 3, no. 2
Influence of Experts’ Domain-specific
Knowledge on Risk Taking in Adversarial
Situations
Philippe Chassy1 and Fernand Gobet2
1 Mathematical Psychology lab, University of Liverpool, UK
2 London School of Economics and Political Science, UK
Correspondence: Philippe Chassy, Philippe.Chassy@liverpool.ac.uk
Abstract
Experts play a considerable role in society, as they have to evaluate the risk of policies in many fields of
social life and in adversarial situations (e.g., the military). Yet, the influence of expertise on risk taking
in adversarial situations has received little attention. An examination of the strategies used by chess
players ranging from amateurs to masters in competitive games (n = 73,341) revealed an unexpected
pattern of results. First, the majority of players favored the riskier strategy. This result is in line with the
literature on economic decision making that indicates a tendency to take risks in situations where the
outcome can be either positive or negative. More surprising is our second finding: As skill increased, the
majority of players still adopted a risk seeking attitude but the proportion of players taking a more
conservative approach increased. This result would tend to indicate that experts making decisions with
impact on their own life become increasingly risk averse. Overall, our findings indicate that knowledge
moderates but does not eliminate risk taking behavior. They also highlight that risk taking in adversarial
situations might result from a complex set of factors. Further research should establish which
psychological processes drive players to adopt a risk taking or conservative strategy in their games.
Keywords
Risk, expertise, uncertainty, strategy, adversarial situations
Introduction
Evolutionary pressures drive individuals to
compete for resources. Occasionally, people or
groups decide to engage in high-risk operations to
win over the competition. The confrontation,
whether direct or indirect, will have significant
consequences for all parties involved. At the
group level, high-risk operations are common in
institutionalized activities such as politics,
business, and the military. Competition can be
observed at the individual level in both direct
confrontations such as in sports and indirect
oppositions such as promotion at work. In these
real-life situations, characterized by potential loss,
people face uncertain environments where risks
cannot be evaluated precisely. Due to this
uncertainty, society relies on experts to inform
and often establish strategies (Knighton, 2004;
Vertzberger, 1995). Despite its importance, the
actual influence of experts on risk taking in high-
risk situations has not received much attention in
the scientific literature (Gobet, 2016). Yet as
experts inform key decisions in society, there is a
need to understand how experts evaluate risk and
set their attitude with respect to potential losses.
In the present paper, we evaluate the influence of
expert knowledge on strategic risk by examining
Journal of Expertise
2020. Vol. 3(2)
© 2020. The authors
license this article
under the terms of the
Creative Commons
Attribution 3.0 License.
ISSN 2573-2773
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 89
Journal of Expertise / June 2020 / vol. 3, no.2
decisions at the individual level. To ensure that
decisions are concrete, we selected a domain
where deciders incur the consequences of their
decisions.
Attitude to risk has been extensively explored
in laboratory conditions with experiments
involving economic decisions (Kahneman &
Tversky, 1979; Tversky & Kahneman, 1992). In
situations that can have either a positive or a
negative outcome, the bulk of the evidence points
to individuals being risk seeking (Birnbaum,
2008; Ert & Erev, 2013). Perceived risk has been
shown to depend upon how the decider frames the
situation (Birnbaum, 2008; Kahneman & Tversky,
1984). Crucially, it is during the phase of building
a representation of the situation that experts’
domain-specific knowledge provides them with a
substantial advantage over non-experts. This
representation, built nearly instantaneously
(Chassy & Gobet, 2011a), enables experts to solve
simple problems correctly within a few seconds
(Van der Maas & Wagenmakers, 2005).
Considering that recognition of strategic features
is rapid and immediately offers potential solutions
(Bilalić et al., 2009; Chassy, 2013), the influence
of such domain-specific knowledge should be
sizable on risks estimates. It is known that people
tend to avoid situations that might involve
potential loss even if there is the possibility of a
gain (Ert & Erev 2008). But, once they face the
prospect of a potential loss individuals become
risk seeking (Scholer et al., 2010). The immediate
prospect of the loss promotes risk seeking
behaviora mechanism that is served by specific
brain regions (Tom et al., 2007). The evidence
from various domains of expertise, on the other
hand, indicates that experts tend to perceive less
risk than laypeople when evaluating the level of
risks in potentially damaging technologies (Slovic
et al., 1995). A study from Savadori et al. (2004)
has provided evidence that experts do not base
their judgement on the same cues as laypeople.
Taken together, these studies indicate that
domain-specific knowledge modifies the
construction of the problem situation, and by
doing so impacts on the final estimate of risk.
These studies, however, do not put the expert
in a situation wherein they would incur the
consequences of their decisions. In addition, and
crucially, they refer to hypothetical situations,
where experts advise a policy on a fictitious
experimental scenario. Testing risk taking in high-
risk confrontations requires meeting several
methodological criteria: a real-life situation to
ensure ecological validity of the findings, a
domain of expertise that puts the experts in a
position wherein risk is quantifiable, a situation
that is competitive, and, crucially, decisions that
will impact on the decider.
Chess is one of the very few domains that
meets all these methodological criteria. It is a
zero-sum game where the gain of one player
corresponds to the loss of the other player.
Players’ skill level is measured precisely and
quantitatively by the Elo rating system (Elo,
1978). This feature, which is nearly unique in
research into expertise, is particularly useful as, in
adversarial situations, the opponent’s actions
might affect the course of events and with it the
outcome. Knowing the opponent’s level of
expertise provides important information when
framing the situation. Three additional features
contribute to the use of chess as a fitting domain
for studying risk. First, it is a visuospatial game
played with the same rules worldwide, which
avoids biases linked to language (Casasanto,
2008). Second, the presence of databases
including games played all over the world ensures
that the results will not be biased by cultural
approaches to risk (e.g., Li et al., 2009). Last and
most importantly, the Elo rating, initially
developed as a measure of expertise, is regarded
by chess players as a reward system. For most if
not all players, Elo points are at the core of their
chess life, as they not only indicate the level of
expertise (a source of prestige) but also are the
key factor upon which players are invited to play
in tournaments, to participate in team
competitions, and to deliver lectures.
In this context, the literature indicating
dominance of loss aversion in mixed gambles (Ert
& Erev, 2008; Scholer et al., 2010) implies that
experts take the risky option to avoid high losses
when they face weaker opponents. The attitude of
the weaker players is subtler. The larger the rating
difference with the better player, the higher the
gain in case of a win but also the higher the
probability of losing small. If the weaker player
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 90
Journal of Expertise / June 2020 / vol. 3, no.2
focuses on the potential gain (winning a large
number of Elo points), then a risk-taking strategy
might be undertaken; however, if the player
focuses on the low probability of winning, then a
more conservative strategy might be undertaken.
To understand which factor is the most important
to non-experts, Slezak and Sigman (2012)
examined the speed-accuracy tradeoff of non-
expert players (rating < 2000 Elo) in games where
the thinking time was limited to three minutes for
each player. The results indicated that weaker
players tended to focus on accuracy by playing
slower. Players are thus sensitive to risk,
becoming more conservative when meeting better
rated opponents.
Our study aimed at expanding on this initial
finding by examining the decisions of experts. We
examined the strategies chosen by chess players
during competitive events that took place during
one entire year all over the world. Our analysis is
based on the first move played in the game, a
move that is often chosen by players before the
game. At the beginning of the game, there is a
definite number of strategies to choose from
(Matanović et al., 1971). Different strategies
involve different levels of risk. The fact that
players decide the strategy they are going to play
beforehand is demonstrated by the fact that they
know the first moves by rote memory (Chassy &
Gobet, 2011b) and that they can identify the
strategy upon mere recognition of the position
(Chassy, 2013).
Based on the reviewed literature, we put
forward three hypotheses. First, considering that
chess players willingly engage in a game which in
essence is a battle, we would expect the players to
be predominantly risk takers in this context.
Second, the theoretical considerations discussed
above suggest that experts’ domain-specific
knowledge will enable them to generate a more
accurate representation of the situation and thus to
evaluate risk better; thus, based on the fact that all
players are essentially sensation seekers (Joireman
et al., 2002), we predicted that experts would be
more risk taking than non-experts. Third, we
would predict that players adapt to their opponent
levels of skill so as to minimize loss.
Method
Materials
We used the games from a commercial chess
database (Fritz Database, version 12; Morsch,
2009). We selected all the games played over a
year, namely from April 1, 2008, to March 31,
2009, by players whose rating spanned the
1600-2399 Elo range. The final sample
consisted of 73,341 games. Players were
assigned to one of four groups: amateurs (1600-
1799 Elo, M = 1719.18; SD = 56.57), club
players (1800-1999 Elo, M = 1915.62;
SD = 56.05), candidate masters (2000-2199 Elo,
M = 2103.34; SD = 57.21), or masters (2200-
2399 Elo, M = 2292.64; SD = 57.16). (To
minimize ambiguity between the candidate-
master and the master classes, we will refer to
the candidate-master class as the candidate
class.) The cutoff point at 2,000 Elo corresponds
to the definition of chess expertise in the
scientific literature (Elo, 1978). Hence, our
sample was made of two non-expert and two
expert groups. The four selected skill levels
ensured that a sufficiently large number of
games were used and had the advantage that the
levels occupied adjacent positions in the rating
scale, which made comparisons easier.
Measure of Risk
Although other measures exist, it is common in
research on judgment and decision making to
use standard deviation (σ) around the expected
value (μ) to define risk (Rothschild & Stiglitz,
1970; Damodaran, 2007). The application of
this definition to risk taking in chess has already
proven useful to analyze attitude to risk in
adversarial situations (Chassy & Gobet, 2015).
After partitioning the data set as a function of
the first move, we calculated the variance
around the mean outcome for each skill to
determine their level of risk (see Appendix 1).
Based on an extensive chess literature (e.g.,
Matanović, et al., 1971), one can categorize the
first moves of the game in two main groups:
open games (1.e4) and closed games (1.d4, 1.c4,
and 1.Nf3). Prior to our sample year, these four
moves account for 96% of all openings on a
period covering 5 years (2003-2007), for the
same player range (1600-2399). To estimate risk
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 91
Journal of Expertise / June 2020 / vol. 3, no.2
directly from the empirical data, we used the
games in the Fritz database (Big Database
2010). For all games where moves were
traceable during the period of 2002-2007, we
analyzed the pattern of wins, draws, and losses.
The time window 2002-2007 was selected so as
to be close to the period being analyzed and to
be sufficiently large to enable accurate
estimates. Games with only one move or
incomplete games were removed from the
database (n < 1%). Risk, defined as variance
around the mean outcome, was estimated by
analyzing 1,474,378 games. For all skill levels
pooled, σ was 42.69% for open games and
41.59% for closed games; the difference is
statistically highly reliable given the large
number of observations (see below). In line with
the definition of risk, as compared to closed
games open games increase the chances of both
winning (39.58% vs 38.91%) and losing
(33.66% vs 30.93%) while decreasing the
chances of drawing (26.75% vs 30.16%). Our
empirical definition of risk is in line with the
subjective experience of chess authorities for
over a century (e.g., Aagard, 2002; Gunsberg,
1901). Since open games involve more risks,
they were labeled as “risky”, and closed games
were labeled as “conservative”. The reader
should bear in mind the fact that these labels
refer to relative risk, since conservative
openings still entail some degree of risk.
Payoffs
To shed light on the factors influencing the
choice of risky or conservative strategies by
chess players, we calculated the payoff for each
game. We followed the mathematical
procedures developed by Elo (1978). The first
step consisted in computing the rating difference
between the player initiating the game
(henceforth, first player) and the player playing
the second move (henceforth, second player),
see equation 1 below. Then, we calculated the
probability of winning of the first player in each
game (see equation 2). The results were checked
against the table of probabilities provided by the
International Chess Federation (FIDE).
The second step consisted in calculating the
potential change in rating (see equations 3 and 4
below). As indicated by FIDE, the k parameter
is adjusted to various situation (e.g., age of the
player, stage of career). To facilitate
comparisons, we used k = 20 for the four skill
levels. As a control for the correctness of our
calculations of the payoffs and probabilities, we
calculated the utility of each game (Von
Neumann & Morgenstern, 2007). All games
should have a zero value as chess is, by
definition, a zero-sum game; this theoretical
assumption was verified (see Appendix 2).
Rating difference: = Elosecond Elofirst
(1)
Probability of winning of the first player: p = (1+10/400)-1
(2)
Payoff of the game (win): Payoff = k(1 p)
(3)
Payoff of the game (loss): Payoff = k(0 p)
(4)
Results
The data show that the risky strategy was
selected more often (54.40%) than the
conservative strategy (45.60%); a difference that
that is statistically significant as indicated by a
chi-square analysis on frequencies,
χ²(1, n = 73,341) = 569.184, p < .05, φ = .09.
This result supports the hypothesis that chess
players are predominantly risk seeking. To
explore whether the level of expertise influences
risk taking, we compared the frequencies of
occurrence of risky and conservative systems as
a function of skill level. Figure 1 shows that, at
all risk levels, risky strategies dominate
conservative strategies. A key finding is that, as
their skill increases, players tend to be more
conservative. The difference between levels of
expertise is statistically significant,
χ²(3, n = 73,341) = 117.742, p < .05 , φ = .04.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 92
Journal of Expertise / June 2020 / vol. 3, no.2
Figure 1. Percentage use of the risky strategy for each group of skill.
We then analyzed whether players are
influenced by the level of the opponent in their
attitude to risk. This is a crucial analysis since it
shows whether players adapt to the opponent’s
level of skill. Figure 2 shows the percentage of
first players who selected a risky strategy as a
function of the level of skill of their opponents.
While variations in the use of risky strategies
can be noted within a class, Figure 2 confirms a
global trend of decreasing risk as the skill level
of a player increases. But Figure 2 also
highlights two noticeable situations: Amateurs
meeting Masters and vice-versa. Amateurs, like
Masters, turn more cautious than against any
other opposition when they meet the opposite
end of the spectrum of skill.
Figure 2. Risk taking of the first player as a function of the skill level of the first player and the opponent.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 93
Journal of Expertise / June 2020 / vol. 3, no.2
Table 1 reports, for each of the 16 conditions,
the average payoff and probability, the
percentage of risky strategy, and the χ2 test of
independence. Payoff and probability
characterize the situation that players were
facing. Their decision, which is whether to
adopt a conservative or risky strategy, is
reflected in the average percentage of risky
strategies used. To illustrate with an example,
we consider the case of first players at club level
facing a second player at master level. In the
case of a win, the club players could win 17.45
Elo points on average and in the case of a loss,
they incurred an average loss of 2.55 Elo points;
the low probability of winning (p = .13) and the
high probability of losing (p = .87) defined the
challenge these players were facing (see
Appendix 2). In these conditions, club players
decided to use the risky strategy in 55.90% of
the games; a choice that departs significantly
from an equal distribution of conservative and
risky strategies, as indicated by the associated
significant χ² value of 36.82.
Table 1. Mean values for the payoffs and probabilities in case of victory or defeat of the first player.
Player
Victory
Defeat
n
χ²
First
Second
Payoff
p
Payoff
p
Amateur
Amateur
9.84
0.51
-10.16
0.49
994
18.06*
Club
14.84
0.26
-5.16
0.74
1937
52.54*
Candidate
17.55
0.12
-2.45
0.88
1386
23.90*
Masters
19.13
0.04
-0.87
0.96
341
0.66
Club
Amateur
5.10
0.75
-14.90
0.26
2042
40.62*
Club
9.97
0.50
-10.03
0.50
3522
47.73*
Candidate
14.66
0.27
-5.34
0.73
6159
128.32*
Masters
17.45
0.13
-2.55
0.87
2644
36.82*
Candidate
Amateur
2.47
0.88
-17.53
0.12
1466
22.10*
Club
5.30
0.74
-14.70
0.27
6455
84.60*
Candidate
9.97
0.50
-10.03
0.50
9561
101.89*
Masters
14.57
0.27
-5.43
0.73
10352
100.52*
Masters
Amateur
0.87
0.96
-19.13
0.04
363
0.22
Club
2.55
0.87
-17.45
0.13
2702
8.33*
Candidate
5.39
0.73
-14.61
0.27
10737
21.37*
Masters
10.00
0.50
-10.00
0.50
12680
9.55*
Note: * Significant at p < .05, n: number of games
The independence tests reported in Table 1
indicate for each of the 16 conditions whether the
distribution of games using risky or conservative
strategy departs from chance or not. In fourteen
conditions out of 16, chess players significantly
favored the risky strategy over the conservative
one. Only when amateurs and masters met did the
players not show this preference. As indicated in
Table 1, the significant differences are found in
the two conditions that have the lower number
of cases, casting a shadow on the validity of the
finding.
Discussion
This paper has investigated risk taking in chess,
a zero-sum confrontational situation. We have
analyzed the strategic choices made by players
of four different levels of skill (amateur, club,
candidate, and master level). The analysis was
conducted on 73,341 chess games that were
played across the globe in an entire year. The
results support the hypothesis that chess players,
whether expert or not, tend in general to choose
the riskier option at the beginning of the game.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 94
Journal of Expertise / June 2020 / vol. 3, no.2
The second result of interest is that risk levels
are inversely proportional to skill. Finally, due
to differences in statistical power across
conditions, we cannot come to a definite
conclusion as to whether our data support our
third hypothesis; it remains unclear whether
players adapt to the opponent. Our findings
inform the current literature on risk taking and
open new avenues for investigating decision
making processes in adversarial situations.
As predicted by Tversky and Kahneman’s
(1992) theory of decision making, loss aversion
drove a majority of players to use more risk-
seeking strategies. Yet, a substantial 45.60% of
the players favored conservative strategies, a
figure highlighting the specificity of
confrontational situations. The dynamics
underlying the choice of a strategy, whether
conservative or risky, emerge from interaction
between numerous factors. One factor that
might play a key role is players’ personality. It
is known that children who play chess score
higher on the Big Five dimensions of
Intellect/openness and Energy/extraversion than
children who do not play chess (Bilalić et al.,
2007), although this pattern of results was not
found with adult chess players (Vollstädt-Klein
et al., 2010). Another personality trait
potentially playing a role in determining the
level of risk undertaken by players is sensation
seeking (Horvath & Zuckerman, 1993)the
keenness to engage in activities yielding intense
experiences. Chess players have been shown to
score higher than the general population on
sensation seeking (Joireman et al., 2002), a trait
that is unsurprisingly correlated with risk taking
behaviors (Kern et al., 2014). We speculate that
chess players’ attitude to risk results from the
balance between the natural tendency to avoid
loss, on the one hand, and chess players’
personalities, on the other hand. The net result
of these opposing forces, which vary from
player to player, determines whether a player
will be driven primarily by loss aversion and
choose the conservative option or by sensation
seeking and thus be more risk taking.
Our results highlight that experts use more
conservative strategies. This key result is
unexpected since it stands in stark contrast with
previous literature. In domains of expertise that
do not include competition or confrontations,
estimates of risk by experts have been lower
than estimates of risk by laypeople (Slovic, et
al., 1995). Since experts perceive less risk in a
given situation, they usually are more risk
taking than non-experts who are more
conservative because of the higher level of
perceived risk. Assuming that experts have on
average the same level of risk tolerance would
lead naturally to the conclusion that they might
take more risks. Another factor that might
increase risk taking in confrontational situations
is self-confidence (Krueger & Dickson, 1994).
Since experts in various domains demonstrate
high levels of confidence (Shanteau, 1988), we
could reasonably think that the level of risk
would be higher than for laypeople. Our results
suggest otherwise. It appears that, in
confrontational situations, loss aversion
outweighs self-confidence and risk seeking, and
this is more the case with experts than with club
players. We would attribute this counter-
intuitive result to the fact that our study is
strictly ecological. Most studies conducted in
the field of risk and risk taking are laboratory
manipulations of fictitious costs and gains.
Although these studies are informative about the
cognitive processes underpinning the evaluation
of costs, the deciders do not incur any real
penalty in case of losses. The chess players in
our sample were putting their Elo rating at stake
in each game. The reality of a potential loss has
made chess players relatively conservative, thus
showing sensitivity to potential loss.
Paradoxically, chess players’ conservatism
might directly stem from their huge amount of
domain-specific knowledge. The representation
of the situation, which is richer for experts
(Campitelli & Gobet, 2004), drives them to
consider more potential outcomes and thus get a
more accurate representation of uncertainty; this
in turn might generate more loss aversion. A
second factor potentially accounting for the
conservative attitude of many players is the fact
that the situation is confrontational. While most
experiments conducted on risk-taking provide
probabilities of losing, they are set as gambles
where the loss of the decider is not necessarily
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 95
Journal of Expertise / June 2020 / vol. 3, no.2
beneficial to a competitor. In a chess game,
there is a will of the opponent to win, and thus if
nothing is done, the game will be lost. As it
entails a threat, this confrontational situation
might make players more prudent.
Our third hypothesis stating that players
adapt risk to the opponent’s skill level has found
some supportive evidence in the statistical
analysis of the distribution of risky and
conservative strategies; however, as highlighted
above, in the two cases where statistically
significant differences were not observed, the
number of players was drastically smaller than
in the other fourteen conditions. Thus, there is
an issue of statistical power. On the one hand,
statistics do not disprove our views. There is
some, even if fractional, evidence that amateurs
and master players adapt to their opponent when
the Elo rating difference is at its highest.
However, the analysis relies on a relatively
limited number of cases as compared to other
conditions. In several of the other conditions,
the same proportion of conservative strategies
revealed a statistical difference between
conservative and risky strategies favoring the
latter ones. So, on the other hand, we have not
reached the point where evidence is
undisputable. As a consequence, it is not clear
whether there is a definite trend or whether our
data are an artefact reflecting random
fluctuations in the use of conservative strategies.
We have to conclude at this point that the debate
remains open and must be investigated in future
studies.
The present study has limitations that should
be kept in mind for a correct interpretation of
the results. An important aspect of the study was
that damage was limited to chess rating. Even
though Elo ratings impact on status and
potential income, they are definitely different
from physical damage and our results are not
predictive, for example, of the levels of risk in
combat situations. Second, the effects sizes are
small; thus, while the results are theoretically
important, their practical implications might be
more limited. A third factor to bear in mind is
that the culture of the domain plays a role in
biasing non-experts’ risk attitudes. In chess,
many chess books value openings that lead to
aggressive, high-risk situations for the first
player (e.g., Levy & Keene, 1976). The first
move 1.e4, which was found risk-seeking in our
study, was even promoted by a world champion
as the best move (Fischer, 1995). This type of
advice and the social pressure that might
accompany it could bias non-experts to take
more risks. As noted by Pleskac and Hertwig
(2014), social norms can influence risk taking,
and chess is certainly a domain where the norm
is to take risks. Fourth, though competitive, non-
expert players might approach the game paying
less attention to the potential outcome, as it
would not impact them as much as it would
impact professional players. Lastly, we would
like to mention that our measure of risk based
on the first move is somewhat arbitrary. Our use
of the first move has the merit of measuring risk
ecologically, but players’ attitude to risk might
be more subtle than a binary choice between
risky and conservative options.
A new picture of risk taking in adversarial
situations emerges. Whereas, in line with
theoretical predictions, chess players are
predominantly risk taking when playing chess,
the results also brought to light the fact that a
significant proportion of players chose the more
conservative option. Such a finding calls for a
revision of risk-taking theories that predict
unconditional loss-averse behaviors. With
respect to the influence of domain-specific
knowledge, we have showed that experts in
adversarial situations are more conservative,
thus suggesting that domain-specific knowledge
increases loss aversion. The extent to which this
finding applies to other adversarial situations is
left undetermined but opens an avenue for
future investigations on the relationship between
domain-specific knowledge and attitude to risk.
In conclusion, experts’ strategic thinking
integrates two factors: domain-specific
information which makes experts more
conservative and the level of expertise of the
opponent which can make experts more risk
taking.
Author’s Declarations
The authors declare that there are no personal or
financial conflicts of interest regarding the
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 96
Journal of Expertise / June 2020 / vol. 3, no.2
research in this article.
The authors declare that they conducted the
research reported in this article in accordance with
the Ethical Principlesof the Journal of Expertise.
The authors declare that they are not able to make
the dataset publicly available but are able to
provide it upon request.
References
Aagard, J. (2002). Excelling at chess. London:
Everyman Chess.
Bilalić, M., McLeod, P., & Gobet, F. (2007).
Personality profiles of young chess players.
Personality and Individual Differences, 42,
901-910.
Bilalić, M., McLeod, P., & Gobet , F. (2009).
Specialization effect and its influence on
memory and problem solving in expert chess
players. Cognitive Science, 33, 1117-1143.
Birnbaum, M. H. (2008). New paradoxes of
risky decision making. Psychological
Review, 115, 463-501.
Campitelli, G., & Gobet, F. (2004). Adaptive
expert decision making: Skilled chessplayers
search more and deeper. Journal of the
International Computer Games Association,
27, 209-216.
Casasanto, D. (2008). Who's afraid of the big
bad Whorf? Crosslinguistic differences in
temporal language and thought. Language
Learning, 58, 63-79.
Chassy, P. (2013). The role of memory
templates in experts’ strategic thinking.
Psychology Research, 3, 276-289.
Chassy, P., & Gobet, F. (2011a). A hypothesis
about the biological basis of expert intuition.
Review of General Psychology, 15, 198-212.
Chassy, P., & Gobet, F. (2011b). Measuring
chess experts' single-use sequence
knowledge: An archival study of departure
from 'theoretical' openings. PLoS ONE,
6(11), e26692.
Chassy, P., & Gobet, F. (2015). Risk taking in
adversarial situations: Civilization
differences in chess experts. Cognition, 141,
36-40.
Damodaran, A. (2007). Strategic risk taking: A
framework for risk management. Upper
Saddle River, NJ: Wharton School
Publishing.
Elo, A. (1978). The rating of chessplayers, past
and present. New York: Arco.
Ert, E., & Erev, I. (2008). The rejection of
attractive gambles, loss aversion, and the
lemon avoidance heuristic. Journal of
Economic Psychology, 29(5), 715-723.
Ert, E., & Erev, I. (2013). On the descriptive
value of loss aversion in decisions under risk:
Six clarifications. Judgment and Decision
Making, 8, 214-235.
Fischer, R. J. (1995). My 60 memorable games.
London: Batsford.
Gobet, F. (2016). Understanding expertise.
London: Palgrave Macmillan.
Gunsberg, I. (1901). The chess openings.
London: Bell and Sons.
Horvath, P., & Zuckerman, M. (1993).
Sensation seeking, risk appraisal, and risky
behavior. Personality and Individual
Differences, 14, 41-52.
Joireman, J. A., Fick, C. S., & Anderson, J. T. I.
(2002). Sensation seeking and involvement
in chess. Personality and Individual
Differences, 32, 509-515.
Kahneman, D., & Tversky, A. (1979). Prospect
theory: An analysis of decisions under risk.
Econometrica, 47, 263-291.
Kahneman, D., & Tversky, A. (1984). Choices,
values and frames. American Psychologist,
39, 341-350.
Kern, L., Geneeau, A., Laforest, S., Dumas, A.,
Tremblay, B., Goulet, C., et al. (2014). Risk
perception and risk-taking among
skateboarders. Safety Science, 62, 370-375.
Knighton, R. J. (2004). The psychology of risk
and its role in military decision-making.
Defence Studies, 4, 309-334.
Krueger, N., & Dickson, P. R. (1994). How
believing in ourselves increases risk taking:
perceived self efficacy and opportunity
recognition. Decision Sciences, 25, 385-400.
Levy, D., & Keene, R. (1976). An opening
repertoire for the attacking club player.
London: Batsford.
Li, S., Bi, Y.-L., & Zhang, Y. (2009). Asian risk
seeking and overconfidence. Journal of
Applied Social Psychology, 39, 2706-2736.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 97
Journal of Expertise / June 2020 / vol. 3, no.2
Matanović, A., Molorović, M., & Božić, A.
(1971). Classification of chess openings.
Belgrad: Chess Informant.
Morsch, F. (2009). Fritz (Version 12).
Hamburg, Germany: Chessbase.
Pleskac, T. J., & Hertwig, R. (2014).
Ecologically rational choice and the structure
of the environment. Journal of Experimental
Psychology: General, 143(5), 2000-2019.
Rothschild, M., & Stiglitz, J. E. (1970).
Increasing risk: I. A definition. Journal of
Economic Theory, 2(3), 225-243.
Savadori, L., Savio, S., Nicotra, E., Ruminati,
R., Finucane, M., & Slovic, P. (2004). Expert
and public perception of risk from
biotechnology. Risk Analysis, 24, 1289-1299.
Scholer, A. A., Zou, X., Fujita, K., Stroessner,
S. J., & Higgins, E. T. (2010). When risk
seeking becomes a motivational
necessity. Journal of Personality and Social
Psychology, 99(2), 215.
Shanteau, J. (1988). Psychological
characteristics and strategies of expert
decision makers. Acta Psychologica, 68, 203-
215.
Slezak, D. F., & Sigman, M. (2012). Do not fear
your opponent: Suboptimal changes of a
prevention strategy when facing stronger
opponents. Journal of Experimental
Psychology: General, 141(3), 527-538.
Slovic, P., Malmfors, T., Krewski, D., Mertz, C.
K., Neil, N., & Bartlett, S. (1995). Intuitive
toxicology II: Expert and lay judgments of
chemical risks in Canada. Risk Analysis, 15,
661-675.
Tom, S. M., Fox, C. R., Trepel, C., & Poldrack,
R. A. (2007). The neural basis of loss
aversion in decision-making under risk.
Science, 315(5811), 515-518.
Tversky, A., & Kahneman, D. (1992). Advances
in prospect theory: Cumulative
representation of uncertainty. Journal of Risk
and Uncertainty, 5, 297-323.
Van der Maas, H. L. J., & Wagenmakers, E. J.
(2005). A psychometric analysis of chess
expertise. American Journal of Psychology,
118, 29-60.
Vertzberger, Y. Y. L. (1995). Rethinking and
reconceptualizing risk in foreign policy
decision-making: a sociocognitive approach.
Political Psychology, 16, 347-380.
Vollstädt-Klein, S., Grimm, O., Kirsch, P., &
Bilalić, M. (2010). Personality of elite male
and female chess players and its relation to
chess skill. Learning and Individual
Differences, 20, 517-521.
Von Neumann, J., & Morgenstern, O. (2007).
Theory of games and economic behavior.
Princeton University Press.
Submission: 22 March 2018
Resubmission: 25 January 2020
Acceptance: 4 May 2020
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 98
Journal of Expertise / June 2020 / vol. 3, no.2
Appendix 1. Calculation of risk
Risk is calculated as variation around the mean. For a sample of n observations of outcome, risk is calculated
in three steps:
x
̅ = 𝑥
𝑛
σ2 = ∑(x-x
̅)2 / n
Risk = σ2
In full agreement with the regulations of the International Chess Federation (Section 11.1 of the FIDE
Laws of Chess [see https://www.fide.com/FIDE/handbook/LawsOfChess.pdf]), a win is coded as 1, a loss as
0 and a draw as 0.5. Following the same regulations, the expected result is the win probability plus half the
probability of drawing.
Hence an opening move that would lead to 40% wins, 30% draws and 30% losses on a sample of
n = 1,000 games would have a mean expectancy of
M = (.40 1) + (.30 1/2) = .55 thus entailing a risk variance of
σ2R = [(1 .55)2 400 + (.5 .55)2 300 + (0 .55)2 300)] / 1000 = 0.1725 and so,
risk = 0.1725 = 0.415331.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 99
Journal of Expertise / June 2020 / vol. 3, no.2
Appendix 2: Zero-Sum Game and Distribution of Payoffs in the Sample
Chess is a zero-sum game, so the gain of one player is the loss of the other. For example, let us consider
player A, rated 1800 Elo, playing against player B, rated 2200 Elo. By following the formulas provided
by Elo (1978) and rounding numbers to the nearest integer, we obtain the following information: Player
A has a 9% chance of winning 18 Elo and a 91% chance of losing 2 Elo. In case of a win, Player A wins
18 Elo and Player B loses 18 Elo. In case of loss, Player A loses 2 Elo and Player B wins 2 Elo. In both
cases, the loss of a player equals the gain of the other, but the stakes differ according to the outcome of
the game.
In our sample of 73,431 games, the combination of four levels of skill for the first player and four
levels of skill of the second player generates 16 conditions. The probabilities of winning and losing for
the first player are depicted in Figure A1. The probability distribution of winning as calculated from the
data is in dark green and the probability distribution of losing in light green. Figure A1 shows that the
probabilities for the win and loss prospects are mirror images of one another as they naturally sum to 1.
Figure A2 depicts the distribution of payoff in Elo points in case of gains (dark blue) and loss (light
blue). Density for payoffs exactly shows the distribution of potential gains and losses for all 73,341
games. Taken together, Figures A1 and A2 depict the prospects that the players were facing.
In all games, following the formulas provided by Elo (1978), we found that the added expected
utilities for each outcome cancelled each other, confirming that chess is a zero-sum game.
Figure A1. Probability distribution of winning (dark green) and losing (light green) for the first player.
Chassy and Gobet (2020) Expertise and Risk in Adversarial Situations
https://www.journalofexpertise.org 100
Journal of Expertise / June 2020 / vol. 3, no.2
Figure A2. Distribution of payoffs as a function of the skill of the first and second players
for both winning (dark blue) and losing (light blue), from the first player standpoint.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Four studies demonstrate the importance of self-regulatory mechanisms for understanding risk-seeking behavior under loss. Findings suggest that risk seeking becomes a motivational necessity under 3 conditions: (a) when an individual is in a state of loss; (b) when the individual is in a prevention-focused regulatory state (E. T. Higgins, 1997); and (c) when the risky option alone offers the possibility of eliminating loss. In situations involving loss, prevention motivation but not promotion motivation (whether measured or manipulated) was uniquely associated with behaviors that served the motivation to maintain the status quo. When the risky option offered the sole possibility of returning to the status quo, prevention motivation predicted increased risk seeking. However, when a more conservative option was available that also offered the possibility to return to the status quo, prevention motivation predicted risk aversion. When neither option offered the possibility to return to the status quo, prevention motivation was not associated with risky choice. The authors discuss the benefits of complementing existing accounts of risky decision making under loss with regulatory focus motivational mechanisms.
Article
Full-text available
Previous studies of loss aversion in decisions under risk have led to mixed results. Losses appear to loom larger than gains in some settings, but not in others. The current paper clarifies these results by highlighting six experimental manipulations that tend to increase the likelihood of the behavior predicted by loss aversion. These manipulations include: (1) framing of the safe alternative as the status quo; (2) ensuring that the choice pattern predicted by loss aversion maximizes the probability of positive (rather than zero or negative) outcomes; (3) the use of high nominal (numerical) payoffs; (4) the use of high stakes; (5) the inclusion of highly attractive risky prospects that creates a contrast effect; and (6) the use of long experiments in which no feedback is provided and in which the computation of the expected values is difficult. In addition, the results suggest the possibility of learning in the absence of feedback: The tendency to select simple strategies, like “maximize the worst outcome” which implies “loss aversion”, increases when this behavior is not costly. Theoretical and practical implications are discussed.
Article
Full-text available
In life, risk is reward and vice versa. Unfortunately, the big rewards people desire are relatively unlikely to occur. This relationship between risk and reward or probabilities and payoffs seems obvious to the financial community and to laypeople alike. Yet theories of decision making have largely ignored it. We conducted an ecological analysis of life's gambles, ranging from the domains of roulette and life insurance to scientific publications and artificial insemination. Across all domains, payoffs and probabilities proved intimately tied, with payoff magnitudes signaling their probabilities. In some cases, the constraints of the market result in these two core elements of choice being related via a power function; in other cases, other factors such as social norms appear to produce the inverse relationship between risks and rewards. We offer evidence that decision makers exploit this relationship in the form of a heuristic-the risk-reward heuristic-to infer the probability of a payoff during decisions under uncertainty. We demonstrate how the heuristic can help explain observed ambiguity aversion. We further show how this ecological relationship can inform other aspects of decision making, particularly the approach of using monetary lotteries to study choice under risk and uncertainty. Taken together, these findings suggest that theories of decision making need to model not only the decision process but also the environment to which the process is adapted. (PsycINFO Database Record (c) 2014 APA, all rights reserved).
Article
Full-text available
Skateboarding is considered to be a high risk activity. Although many studies have identified risk factors associated with skateboarding injuries, few have provided detailed in-depth knowledge on participants’ psychological dispositions towards risk behaviors. The aim of this study was to identify individual factors associated with risk perception and risk-taking among skateboarders. Telephone interviews were conducted with 158 skateboarders (mean age = 18.1 years) recruited in 11 Montreal skateparks. Age, self-efficacy, previous injuries, fear of being injured, sensation seeking and experience level were all included in two linear regression models that were run for risk perception and risk-taking. Age, experience level, sensation seeking, and risk perception are significant explanatory variables of risk-taking. Results show that sensation seeking was the only significant factor associated with risk perception. These results allow for a better understanding of the behavior of skateboarders, they highlight the importance of impulsive sensation seeking in risk perception as well as risk-taking. This study characterizes skateboarders who take risks and provides additional information on interventions for injury prevention.
Article
We discuss the cognitive and the psy- chophysical determinants of choice in risky and risk- less contexts. The psychophysics of value induce risk aversion in the domain of gains and risk seeking in the domain of losses. The psychophysics of chance induce overweighting of sure things and of improbable events, relative to events of moderate probability. De- cision problems can be described or framed in multiple ways that give rise to different preferences, contrary to the invariance criterion of rational choice. The pro- cess of mental accounting, in which people organize the outcomes of transactions, explains some anomalies of consumer behavior. In particular, the acceptability of an option can depend on whether a negative outcome is evaluated as a cost or as an uncompensated loss. The relation between decision values and experience values is discussed. Making decisions is like speaking prose—people do it all the time, knowingly or unknowingly. It is hardly surprising, then, that the topic of decision making is shared by many disciplines, from mathematics and statistics, through economics and political science, to sociology and psychology. The study of decisions ad- dresses both normative and descriptive questions. The normative analysis is concerned with the nature of rationality and the logic of decision making. The de- scriptive analysis, in contrast, is concerned with peo- ple's beliefs and preferences as they are, not as they should be. The tension between normative and de- scriptive considerations characterizes much of the study of judgment and choice. Analyses of decision making commonly distin- guish risky and riskless choices. The paradigmatic example of decision under risk is the acceptability of a gamble that yields monetary outcomes with specified probabilities. A typical riskless decision concerns the acceptability of a transaction in which a good or a service is exchanged for money or labor. In the first part of this article we present an analysis of the cog- nitive and psychophysical factors that determine the value of risky prospects. In the second part we extend this analysis to transactions and trades. Risky Choice Risky choices, such as whether or not to take an umbrella and whether or not to go to war, are made without advance knowledge of their consequences. Because the consequences of such actions depend on uncertain events such as the weather or the opponent's resolve, the choice of an act may be construed as the acceptance of a gamble that can yield various out- comes with different probabilities. It is therefore nat- ural that the study of decision making under risk has focused on choices between simple gambles with monetary outcomes and specified probabilities, in the hope that these simple problems will reveal basic at- titudes toward risk and value. We shall sketch an approach to risky choice that
Article
This essay calls both for much greater attention to a relatively neglected but critical issue-area in foreign policy analysis, that of risk, and for a departure from a rational choice approach in the direction of a sociocognitive approach to risk. The sociocognitive approach to risk and risk-taking introduced here represents a substantial departure from the rational choice approach that has dominated the study of risk in foreign policy decision-making. The essay redefines and reconceptualizes the nature of risk in international politics in a manner that is in our view more consistent with actual decision-makers' behavior. Unlike the "black boxing" of the concept of risk that is evident in the rational choice approach to studies of risk, this study opens the "black box" by first disaggregating it into actual risk, perceived risk, and acceptable risk-and then by detailing the attributes of risk. This, we believe, is an essential first step toward a more realistic, accurate, and policy-relevant analysis. The approach presented here views risk as a social construct. Therefore, in explaining diversity in risk preferences, we focus on the taste for risk as a source of variety in risk preference across individuals, and on the process of problem-framing and communication. How risks are framed, especially when problems are ill-defined, becomes highly consequential for how decision-makers will understand and respond to it, and how effectively risk will be communicated both within the decision-making system as well as to outsiders.
Article
The present study examined the relationship between scores on the Sensation Seeking Scale [SSS; Zuckerman, M. (1979). Sensation seeking; beyond the optimal level of arousal. Hillsdale, NJ: Erlbaum.] and involvement in chess within a sample of 112 college students. Students who reported having played chess, and those with more chess experience, evidenced higher scores on both the Total SSS and the Thrill and Adventure Seeking (TAS) subscale, effects which were independent of gender. Higher scores on Disinhibition were also associated with greater chess experience. The present results provide further support for the validity of the TAS scale with regard to involvement in sports, and suggest that more attention be directed to the link between sensation seeking and involvement in low-risk, but theoretically relevant, sporting activities.
Book
"This is the classic work upon which modern-day game theory is based. What began more than sixty years ago as a modest proposal that a mathematician and an economist write a short paper together blossomed, in 1944, when Princeton University Press published Theory of Games and Economic Behavior. In it, John von Neumann and Oskar Morgenstern conceived a groundbreaking mathematical theory of economic and social organization, based on a theory of games of strategy. Not only would this revolutionize economics, but the entirely new field of scientific inquiry it yielded--game theory--has since been widely used to analyze a host of real-world phenomena from arms races to optimal policy choices of presidential candidates, from vaccination policy to major league baseball salary negotiations. And it is today established throughout both the social sciences and a wide range of other sciences. This sixtieth anniversary edition includes not only the original text but also an introduction by Harold Kuhn, an afterword by Ariel Rubinstein, and reviews and articles on the book that appeared at the time of its original publication in the New York Times, tthe American Economic Review, and a variety of other publications. Together, these writings provide readers a matchless opportunity to more fully appreciate a work whose influence will yet resound for generations to come.