Anticipation and Decision Making in Sport

A. Mark Williams, Robin Jackson, A. Mark Williams, Robin C. Jackson, Robin Jackson, A. Mark Williams, Robin C. Jackson

• Publisher: Routledge
• Year: 2019
• eBook ISBN: 9781351380188
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

The ability to anticipate and make accurate decisions in a timely manner is fundamental to high-level performance in sport. This is the first book to identify the underlying science behind anticipation and decision making in sport, enhancing our scientific understanding of these phenomena and helping practitioners to develop interventions to facilitate the more rapid acquisition of the perceptual-cognitive skills that underpin these judgements.

Adopting a multidisciplinary approach — encompassing research from psychology, biomechanics, neuroscience, physiology, computing science, and performance analysis — the book is divided into three sections. The first section provides a comprehensive analysis of the processes and mechanisms underpinning anticipation and skilled perception in sport. In the second section, the focus shifts towards exploring the science of decision making in sport. The final section is more applied, outlining how the key skills that impact on anticipation and decision making may be facilitated through various training interventions.

With chapters written by leading experts from a vast range of countries and continents, no other book offers such a synthesis of the historical development of the field, contemporary research, and future areas for investigation in anticipation and decision making in sport. This is a fascinating and important text for students and researchers in sport psychology, skill acquisition, expert performance, motor learning, motor behaviour, and coaching science, as well as practicing coaches from any sport.

PART I

Characteristics of expert anticipation in sport

1

Postural cues, biological motion perception, and anticipation in sport

Nicholas J. Smeeton, Stefanie Hüttermann and A. Mark Williams

Introduction

It is well established that the ability to anticipate what will happen next is an important component of successful performance in many sports, particularly racket sports and team ball games (Williams, Ford, Hodges, & Ward, 2018). The need to anticipate arises due to the significant spatial and temporal constraints that exist in many sports, necessitating that information be processed in relatively short periods of time so as to enable athletes to plan and execute a timely response to their opponents’ actions. Some examples of situations where anticipation is important include the penalty kick in soccer, the tennis serve, and batting in baseball or cricket. The initial work on anticipation focussed almost exclusively on the ability of athletes to pick up early or advance, most notably postural, cues from an opponent’s movements (i.e. information arising from an opponent’s biological motion) ahead of a key event, such as when a player’s foot or racket makes contact with the ball (Williams, Fawver, Broadbent, Murphy, & Ward, in press). However, recently researchers have reported that anticipation is dependent on the ability to process a range of different sources of information interactively and dynamically as the action unfolds (e.g. pattern recognition, context, probabilistic information) as opposed to relying exclusive on the pickup of postural cues (Williams, Murphy, Broadbent, & Janelle, 2019). Yet it remains apparent that the pickup of postural cues can help guide anticipation either independent of, or in conjunction with, other perceptual-cognitive skills. At the very least, the pickup of postural cues may confirm or reject the accuracy of any initial judgements made based on other sources of information (Cañal-Bruland & Mann, 2015; Triolet, Benguigui, Le Runigo, & Williams, 2013).

In this chapter, we review more than four decades of research that has focussed on identifying the postural cues that are crucial to anticipation in sport. The main objective is to highlight the various approaches that have been used to examine postural cue usage in sport. Generally, the aim in this body of research has been to identify what cues are picked up, when these sources of information become available, and how such information is processed. In the opening section, we review research using traditional methods such as film-based temporal and spatial occlusion, point-light displays, and liquid crystal occlusion glasses to identify both when and what information is picked up during anticipation. The conceptual approach in early studies was heavily grounded in cognitive psychology using terms such as cue usage and information extraction. Next, we review a more recent body of research, grounded in ecological psychology and dynamical systems theory, that makes use of sophisticated data analysis and modelling methods, such as principal component analysis (PCA) to better identify how information is perceived as well as when and what information sources are important. In this latter approach, anticipation is not constrained by a single cue, or even a collection of cues, but rather is an emergent process as athletes becoming attuned to the biological motions emerging from the opponent’s actions. Finally, we propose that a more detailed understanding of the biomechanical constraints imposed on the actions to be anticipated will reveal greater insight into what information is perceived and offer suggestions for future research in this area of study. In order to further delimit the scope of the chapter, we do not review studies in which researchers have attempted to identify the information sources underpinning anticipation using eye movement registration systems, nor do we consider the implications of the findings reported for perceptual-cognitive training. Similarly, the influence of attempts by opponents to disguise key cues or deceive the observer is not directly discussed in this chapter.

Identifying when key cues are picked up: the temporal occlusion paradigm

One of the earliest published reports focussing on the importance of anticipation in sport was carried out in tennis using the film-based temporal occlusion method (Jones & Miles, 1978). The authors presented participants with filmed images of an opponent serving in tennis from a first-person perspective. Participants viewed each serve and were required to anticipate where on the court the opponent was going to play the ball by marking a response on a scaled, schematic representation of the court. At various time points relative to ball-racket contact, the film was occluded in order to prevent access to later arising sources of information. The footage was occluded either 42 ms before, at, or 335 ms after the racket made contact with the ball. Participants were constrained to make a judgement based on information available up to the point of occlusion, which in the earlier occlusion conditions meant that decisions had to be made based on cues emerging from the server in advance of ball flight information. A group of expert tennis players was much more accurate in making anticipation judgements when compared to another group of less expert players, with accuracy scores significantly above chance in the former group, even in the earliest occlusion condition. The expert tennis players were able to pick up postural cues from the opponent ahead of ball contact in order to enable them to successfully anticipate where the ball was likely to land on their side of the court.

This seminal finding led to a plethora of related research using the film-based temporal occlusion paradigm. The generalisability of the findings was explored across a range of different sports, including squash (Abernethy, 1990), badminton (Abernethy & Zawi, 2007), baseball (Moore & Müller, 2014), cricket (Abernethy & Russell, 1984), soccer (Williams & Burwitz, 1993), volleyball (Wright, Pleasants, & Gomez-Meza, 1990), and field hockey (Starkes, 1987), to name only a few sports. The ensuing research largely confirmed the importance of being able to pick up advance postural cues when anticipating an opponent’s actions in sport. Although the key time period for information extraction has been shown to be sport- and task-specific, and to interact with the availability of other sources of information (Roca, Ford, McRobert, & Williams, 2013), a consistent observation is that experts are better than non-experts at picking up the key cues underpinning anticipation. Moreover, with increasing levels of expertise, players are able to extract information from progressively earlier stages in the action (Müller, Abernethy, & Farrow, 2006). An illustration of various temporal occlusion points is presented in Figure 1.1.

FIGURE 1.1 Some example frames of video sequences used in the temporal occlusion approach examining anticipation at the penalty kick in soccer. Milliseconds to and from ball-foot contact are indicated at the top. Initial frame to final frame (left to right) of stimulus videos are presented for temporal occlusion conditions. Final frames represent example occlusion points of a kick.

Perhaps the most notable progress made in the decades that followed was refinement of the methods used to explore research questions. Advances in technology enabled 16 mm film to be replaced with higher resolution digital video footage, and the advent of digital editing software allowed for easier and more accurate editing (Williams, Davids, & Williams, 1999). A shift occurred away from using pen-and-paper responses and towards more realistic movement-based response modes, often incorporating sophisticated measurement devices, such as infrared beams, pressure-sensitive floor mats, and optoelectronic motion capture systems (Müller & Abernethy, 2006; Oudejans & Coolen, 2003; Starkes, Edwards, Dissanayake, & Dunn, 1995). Similarly, life-size projection screens were more frequently used, as were, more recently, three-dimensional video projectors, igloo-style wrap-around video screens, 360-degree video, and virtual reality (Williams et al., 2019). Also, efforts were made to move outside of the laboratory and explore the phenomenon in situ using, for example, video-based, time-use analysis methods (Abernethy, Gill, Parks, & Packer, 2001; Triolet et al., 2013) and computer-controlled liquid-crystal occlusion goggles (e.g. Farrow & Abernethy, 2003). While such advances in technology inevitably enhanced measurement sensitivity, the overall conclusions remained largely consistent in highlighting the importance of picking up the key postural cues as early as possible in the action.

Identifying what cues are picked up: the spatial occlusion paradigm

A shortcoming with the temporal occlusion approach is its inability to isolate the specific sources of information used during anticipation. In order to identify what cues are important, the temporal occlusion approach has to be combined with event or spatial occlusion, or alternative methods, such as eye movement recording and think-aloud verbal protocols (see Williams & Ericsson, 2005). When employing the spatial occlusion approach, the information sources deemed to be important are either occluded (i.e. masked) or removed, usually for the entire duration of the clip. However, the approach may be used in conjunction with the temporal occlusion paradigm in order to simultaneously examine the ‘what’ and ‘when’ questions. Specific cues thought to be informative, such as the hips of the penalty taker in a soccer penalty kick or the server’s arm and racket in the tennis serve, are removed from view using a mask (e.g. black box) or by digitally editing the video to replace the source with the background so that the source effectively disappears. Any decrement in performance in the occluded condition relative to a non-occluded control suggests that the source of information must play a role in anticipation either on its own or in conjunction with other cues (e.g. Causer, Smeeton, & Williams, 2017; Jackson & Mogan, 2007; Müller et al., 2006; Williams & Davids, 1998). An alternative approach involves presenting a cue (e.g. the hips) in isolation with no additional information presented. If performance scores are above chance when only this single cue is presented, the implication is that this information is sufficient to guide anticipation (e.g. Causer et al., 2017; Müller et al., 2006). Some typical spatial occlusion conditions are presented in Figure 1.2.

FIGURE 1.2 Some example frames of video sequences used in the spatial occlusion approach to examine the cues picked up when anticipating the penalty kick in soccer. Milliseconds to and from ball-foot contact are indicated at the top. Initial frame to final frame (left to right) are presented for conditions where the upper body, head, and hip are occluded (top to bottom rows) during the kick.

While film-based approaches have been most widely used in this field of study, researchers have used point-light and stick-figure displays, which are routinely employed in classical literature on biological motion perception (e.g. Abernethy et al., 2001; Abernethy & Zawi, 2007; Cutting, Proffitt, & Kozlowski, 1978; Shim, Carlton, & Kwon, 2006; Ward, Williams, & Bennett, 2002). In this approach, typically optoelectronic motion capture methods are employed to create point-light displays of sports actions, such as a forehand drive in tennis. These images can then be manipulated in much the same way as film-based occlusion methods, with the overall aim being to identify what cues are important and when they are extracted. The approach has the advantage of being able to remove access to background and structural information, ensuring that only the relative motions between limbs remain in an effort to better isolate the minimal information needed to facilitate skilled perception. A typical point-light image of a tennis player performing a forehand drive shot is presented in Figure 1.3, alongside a stick-figure representation.

Innovative recent methods: principal component analysis

A shortcoming with spatial and temporal occlusion is that they only provide information relating to the key information time windows and body regions underpinning anticipation. Moreover, when researchers use the occlusion approach they have to select a priori the time point or body region thought to be informative. As a result, there is potential for subjective bias in the selection of the regions thought to be informative. In an effort to address these limitations, researchers have more recently used PCA to try and isolate essential movement patterns and examine the value or contribution of the information contained in these patterns to successful anticipation (Bourne, Bennett, Hayes, & Williams, 2011; Diaz, Fajen, & Phillips, 2012; Huys, Smeeton, Hodges, Beek, & Williams, 2008).

FIGURE 1.3 A classical point-light display (right side) image presented next to a stick-figure representation (left side) of the same tennis player stimulus performing a forehand drive shot.

PCA is used on time-series analyses to isolate structures or components that capture a proportion of the total variance in the data set that is orthogonal to the other components (for a tutorial, see Daffertshofer, Lamoth, Meijer, & Beek, 2004). Such an analysis is used to reduce the dimensions of a data set into fewer components or structures. From this analysis, the results are analysed by focussing on: (i) the amount of variance captured by each of the components; (ii) the projection (time evolution) of the components; and (iii) the eigenvector coefficients that provide weightings to the components. The amount of variance captured by each component is ordered or ranked according to the percentage of variance captured. If PCA is successful in reducing the dimensions of the data set, there is a higher than average percentage of variance captured in the principal components, and consequently, the amount of variance captured in the later components is reduced. The projection associated with the component indicates how the data evolve over time. If the components are independent, they will have unrelated time evolutions. Typically, this issue is examined statistically by calculating the covariance between the projections associated with the components. Finally, the eigenvector coefficients indicate the weighting of each time series analysed with PCA on to the component. If a time series contributes a large amount to the component, a high coefficient is reported, whereas a low weighting onto a component is indicated with a low coefficient. These coefficients can be positive or negative, indicating a positive or negative contribution to the component.

PCA has become popular in the motor control literature over recent decades because of its link to contemporary theories of human movement. In particular, it is been used to provide insights into how movements are coordinated and controlled (e.g. see Kelso, 1997; Scholz & Schoner, 1999). It is possible to examine the motions of many body regions and their biomechanical linkages at the same time, consistent with the view that these linkages do not act in isolation (Bernsteĭn, 1967). The potential for data to be separated into movements that have a large amount of coherence and structure over time, and those that have less so i...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication Page
• Table of Contents
• List of figures
• List of tables
• List of contributors
• Preface
• Part I: Characteristics of expert anticipation in sport
• Part II: Characteristics of expert decision making in sport
• Part III: Training anticipation and decision making in sport
• Index