CHAPTER FORTY

The processes underlying game

intelligence skills in soccer players
A. Roca, P. R. Ford and A. M. Williams
Research Institute for Sport and Exercise Sciences,
Liverpool John Moores University, UK
1. INTRODUCTION
The ability of players to anticipate the actions of others and to select appropriate
decisions under time pressure is essential to expert performance in soccer. These
game intelligence skills appear to be dependent on perception and cognition
(Williams et al., 2011). Over the last two decades or so, researchers have sought to
better understand the perceptual-cognitive characteristics that differentiate experts
from less skilled athletes (e.g., see Starkes and Ericsson, 2003). Sophisticated eye
movement registration techniques have provided information on individual and
group differences in visual search behaviours. For example, Williams et al. (1994)
demonstrated that experienced soccer players search the display more extensively
and pick up early arising visual information to guide their performance when
compared with less experienced counterparts.
The majority of researchers studying perceptual-cognitive expertise have
tended to collect eye movement recordings in isolation in order to identify the
sources of information that experts use to guide their performance. There is a lack
of research focusing on how performers process and translate the information
extracted from the visual display into appropriate strategic/tactical decisions.
Verbal protocol analysis techniques can provide this information by assessing the
nature of the cognitive (thought) strategies used during actual performance. Ward
et al. (2003) provided one of the few attempts to record verbal reports during a
film-based recognition task involving soccer. The elite players elicited more
extensive and detailed verbal reports of thoughts when compared with their subelite peers, probably indicating they have more advanced memory representations
(e.g., long-term working memory, LTWM, Ericsson and Kintsch, 1995).
Moreover, although there have been a few successful attempts to identify important
discriminating characteristics between experts and novices in soccer, the majority
of researchers have focused on examining anticipation exclusively, with relatively
few attempts to measure other aspects of performance such as decision making (for
exceptions, see Helsen and Starkes, 1999; Vaeyens et al., 2007). Thus, since

256

Science and Football VII

anticipation and decision making are considered crucial to performance in soccer,

well-controlled efforts are needed to integrate and examine both judgments
concurrently (i.e., what is going to happen next and what is the appropriate
response).
We examine the underlying perceptual-cognitive processes mediating
anticipation and decision making judgements in skilled and less skilled players
during a dynamic, representative soccer task. We expect the skilled players to
outperform their less skilled peers, with systematic between-group differences
being apparent in the underlying perceptual and cognitive processes employed.
2. METHODS
2.1 Participants
A total of 48 male outfield soccer players participated across two experiments. In
Experiment 1, 12 skilled and 12 less skilled players participated, whereas in
Experiment 2, a new set of 12 skilled and 12 less skilled participants took part.
Skilled participants (age 23.5 3.3 years; mean s) were professional and semiprofessional soccer players with an average of 14.5 3.4 years playing experience.
Less skilled participants (age 24.2 3.1 years) were amateur or recreational
players and had taken part in soccer irregularly for an average of 11.2 3.6 years.
2.2 Procedure and task
In both experiments, participants were presented with a representative video-based
task simulation involving four warm-up and 20 test clips of life-size 11 versus 11
dynamic situations filmed from a central defenders perspective. The action
sequences lasted approximately 5 s, with each one being occluded at a key moment
in the action (e.g., player in possession of the ball about to make an attacking pass,
shoot at goal, or maintain possession of the ball by dribbling forward). The test
film was back projected onto a 2.7 m (h) x 3.6 m (w) widescreen using a video
projection system (Hitachi ED-A101, Yokohama, Japan). Participants were
required to take the place of the central defender and move and interact with the
footage as if playing in a competitive match. The accuracy of anticipation and
decision making judgments at occlusion were recorded across the two experiments.
Process tracing measures comprising of visual gaze characteristics
(Experiment 1) and retrospective verbal reports of thoughts (Experiment 2) were
recorded. In Experiment 1, a mobile eye tracking system (Applied Science
Laboratories, Bedford, MA, USA) was used to collect visual search data. This is a
video-based monocular system that measures eye point-of-gaze with respect to a
head-mounted scene camera. In Experiment 2, a wireless microphone system
(Sennheiser EW 112-P G3, Wedemark, Germany) was employed to collect verbal
reports. Participants were given instruction and training on how to think aloud and

Roca et al.

257

provide retrospective verbal reports prior to testing (see Ericsson and Simon,
1993). Each individual training and test session was completed in around 60 min.
2.3 Data analysis
Experiment 1
Anticipation accuracy was defined as whether or not the participant correctly
verbalised the next action of the player in possession of the ball. Decision making
accuracy was defined as whether or not the participant decided on/executed and
verbalised the most appropriate action, which was pre-determined by a panel of
UEFA qualified soccer coaches. Both anticipation and decision making accuracy
were calculated as the mean number of trials (%) in which the participant selected
the correct response. The two measures were analysed separately using
independent t tests.
Two visual search measures were recorded. Search rate data included the
mean fixation duration (in milliseconds), the mean number of fixations, and the
mean number of fixation locations per trial, which were analysed separately using
independent t tests. Percentage viewing time data were analysed using a Group
(skilled, less skilled) x Fixation Location (player in possession of the ball x ball x
opponent x teammate x space) ANOVA. This latter measure was the percentage of
total viewing time spent fixating each of the various areas of the display. Tukeys
post-hoc tests were applied to explore significant interaction effects and a level of
P<.05 was set for statistical significance.
Experiment 2
The measures of anticipation and decision making accuracy were the same as in
Experiment 1. Verbal reports were classified according to a structure adapted from
Ericsson and Simon (1993) and further developed by Ward et al. (2003). Four
types of verbal statement categories were coded: monitoring; evaluations;
predictions; and planning. The mean number of statements per trial in each
category was calculated for both groups. Verbal report data were analysed using a
Group (skilled, less skilled) x Type of Verbal Statement ANOVA. Statistical
significance was set at P<.05.

Science and Football VII

258

3. RESULTS
3.1 Experiment 1
Skilled players were more accurate than less skilled in anticipating the actions of
opponents (68.3 7.2 vs. 37.5 6.5%), t22 = 10.98, P<.001, and in selecting an
appropriate action to execute (80.6 5.2 vs. 49.2 9.6%), t16.71 = 9.96, P<.001.
There were significant skill-based differences in the mean fixation duration,
t22 = -8.24, P<.001, the mean number of fixations, t22 = 9.31, P<.001, and the mean
number of fixation locations per trial, t22 = 9.27, P<.001. Skilled players employed
a more exhaustive search strategy than less skilled counterparts involving more
fixations (12.33 1.79 vs. 6.86 0.97 fixations) of shorter duration (394 83 vs.
766 131 ms) and on significantly more areas of the display (6.67 1.08 vs. 3.47
0.50 fixation locations).
The mean data for percentage viewing time are presented in Figure 1.1. There
was a significant Group x Fixation Location interaction, F2.10,46.29 = 18.90, P<.001.
Post-hoc testing revealed that skilled players spent significantly more time fixating
on the opponents movements and areas of free space in comparison with their
less skilled counterparts. In contrast, less skilled players spent significantly more
time fixating the player in possession of the ball and ball itself compared with the
skilled players.

Space

Less skilled
Skilled

Fixation Location

Teammate
Opponent
Ball
PiP
0

10

20

30

40

50

60

% Viewing Time
Figure 1 Mean

s percentage time spent viewing each fixation location across groups.

(PiP player in possession of the ball).

70

Roca et al.

259

3.2 Experiment 2
Skilled players were more accurate than less skilled counterparts in anticipating the
actions of opponents (69.2 7.7 vs. 34.9 9.0%), t22 = 10.02, P<.001, and in
selecting and executing appropriate tactical decisions (83.0 8.9 vs. 50.6 6.2%),
t22 = 10.32, P<.001.
There was a significant skill-based difference for the total number of verbal
statements generated per trial, F1,22 = 23.18, P<.001. Skilled players (7.58 1.25
statements) verbalised significantly more statements than the less skilled players
(5.25 1.12 statements). The Group x Type of Thought Statement interaction was
not significant, F2.16,47.43 = 1.81, P = .17. The greater number of statements
generated by skilled players compared with the less skilled may have affected this
interaction. Thus, frequency scores were subsequently normalised into proportional
data. Less skilled players were found to make a higher percentage of monitoring
statements compared with the skilled players (67.4 9.8 vs. 47.3 8.4%). In
contrast, skilled players made a higher proportion of evaluation, prediction, and
planning statements than their less skilled peers (52.7 8.4 vs. 32.6 9.8%).
4. DISCUSSION AND CONCLUSION
The skill-based differences in anticipation and decision making judgments were
underpinned by quantitatively different underlying perceptual and cognitive
processes. Skilled players made a greater number of fixations for shorter duration
and towards more informative areas of the display, most likely increasing their
awareness of the positions and movements of opponents/teammates and any
potential areas of free space that may be uncovered or exploited. In contrast, less
skilled players spent longer periods of time watching the ball or the player in
possession of the ball. These data suggest that a more exhaustive search strategy
and analysis of the display appears to be crucial for facilitating superior
anticipation and decision making performance in soccer. Findings support previous
research showing skill-based differences in visual search strategy (Williams et al.,
1994).
Furthermore, when compared with their less skilled counterparts, skilled
players engaged in a greater proportion of evaluation, prediction, and planning
thoughts, suggesting they employed more advanced soccer-specific memory
representations to solve the task-situations. As predicted by Ericsson and Kitschs
(1995) LTWM theory, skilled players likely possess and activate more complex
and superior task-specific memory representations permitting them to easily access
and retrieve specific information of current scenarios and, therefore, facilitate
better anticipation and decision making compared with less skilled peers.
Our findings support and extend previous work (e.g., Williams et al., 2011)
and highlight the perceptual-cognitive processes underpinning superior anticipation
and decision making performance in soccer. The development of these game
intelligence skills may be facilitated through relevant practice and training

260

Science and Football VII

interventions (Stratton et al., 2004). In future researchers should seek to examine

the nature and type of practice (conditions) that lead to the acquisition of superior
judgements and their underlying processes in soccer players. Such research would
provide a principled basis for practitioners involved in talent identification and
development to design the most appropriate learning, training, and practice
environments.
Acknowledgements
The investigation was funded by the Portuguese Foundation for Science and
Technology (FCT) Ministry of Education and Science.
References
Ericsson, K.A. and Kintsch, W., 1995, Long-term working memory. Psychological
Review, 102, 211245.
Ericsson, K.A. and Simon, H.A., 1993, Protocol analysis: Verbal reports as data,
Cambridge, MA: Bradford books/MIT Press.
Helsen, W.F. and Starkes, J.L., 1999, A multidimensional approach to skilled
perception and performance in sport. Applied Cognitive Psychology, 13, 127.
Starkes, J.L. and Ericsson, K.A., 2003, Expert performance in sports: Advances in
research on sport and expertise, Champaign, IL: Human Kinetics.
Stratton, G. et al., 2004. Youth soccer: From science to performance, London:
Routledge.
Vaeyens, R. et al., 2007, Mechanisms underpinning successful decision making in
skilled youth soccer players: An analysis of visual search behaviors. J Motor
Behav, 39, 395408.
Ward, P. et al., 2003, Underlying mechanisms of perceptual-cognitive expertise in
soccer. J Sport Exerc Psychol, 25, S136.
Williams, A.M. et al., 1994, Visual search strategies of experienced and
inexperienced soccer players. Res Quart Exerc Sport, 65, 127135.
Williams, A.M. et al., 2011, Perceptual-cognitive expertise in sport and its
acquisition: Implications for applied cognitive psychology. Applied Cognitive
Psychology, 25, 432442.