Gaze Behaviour of Elite Ice Hockey Players: A Pilot Study

International Journal of Sports Vision, 2001, 7,1,30-36 Gaze Behaviour of Elite Ice Hockey Players: A Pilot Study Joan N Vickers* & Steve.G Martell, University of Calgary, Canada; Mike, Pelino, Team Canada Men s Ice Hockey Team Accepted December 05,2001 Abstract Do elite ice hockey, soccer, and other team sport athlete move their eyes rapidly about the playing environment gaining critical information through rapid fixations of low duration, or alternately, do they exhibit an economy of fixation to fewer locations of longer duration? Research to date in a number of team sports has found conflicting results, with some researchers finding a high frequency/low duration of fixation and others the opposite. In order to explore this problem, the gaze behaviours of Team Canada ice hockey players were recorded as they played defense in a live tactical situation located on the ice. When the tactical problem was situated far from the athlete, then a high frequency and lower duration of fixation was found due to the necessity of scanning a wider field of view. As the opposing players and teammates staed nearer, a lower frequency and longer duration of fixation was found due to the decrease in the distance between opposing players. These results explain, in part, why research defining the gaze of athletes in tactical situations has been equivocal. Both high frequency/low duration and low frequency/high duration fixations occur in team sports, with the guiding principal being the relative position of the athlete to the opponent or teammate in the game setting. Introduction In team sports, information from the playing environment is obtained through all of the senses but primarily through vision. In order for a player to efficiently produce skilled motor movements, he/she must read each situation and react accordingly in a time-pressured environment. The player relies on perceptual and cognitive processes to perceive the environmental situation, make a decision in relation to the play, and execute the appropriate action. Decision-making in many fast-based team sports is therefore reliant on the ability to detect and interpret perceptual information and compare internalized memory structures with situational events (Williams, Davids & Williams, 1999). However, it is unclear if the elite performer s superior attainment is due to the rapid acquisition of visual cues as reported by Williams et al, or the cognitive economy reported by Helsen & Pauwels (1992). These different findings lead to the question - do elite players in soccer, basketball, field hockey, ice hockey and other team sports move their eyes rapidly about the environment gaining critical information through the use of many fixations of low duration, or alternately do they exhibit an economy of fixation to fewer locations for longer durations? *Address for corresponding author: Email: vickers@ucalgary.ca

Visual search techniques have been used to examine the information-processing abilities of expert and non-expert athletes in team and dual sports. In this approach, the eye movements of expert and novice athletes are recorded as they view photographs or video scenes of tactical problems critical to the sport. At the same time, the athlete is required to make a decision or perform other cognitive tasks relative to the scene viewed. Research of this nature has yet to produce a consensus on the perceptual-cognitive aspects of expertise (Abernethy and Russell, 1987; Abernethy, 1990; Helsen & Pauwels, 1992; Williams, Davids & Burwitz, 1994; Williams, Davids, Williams, 1999). For example, Helsen & Pauwels (1992) found that expert soccer players employed an efficient visual search process, consisting of a lower frequency of fixations of longer duration than non-experts. Presumably, this economical process assisted in locating and processing relevant information used in decision-making. On the other hand, Williams et al (1994) and Williams and Davids (1998) investigated the gaze behaviour of elite soccer players in 1 on 1, 3 on 3, and 11 on 11 soccer tactical situations and found that elite players in 11 on 11 situations had a higher frequency of fixation and lower relative duration than non-experts, while in the 1 vs 1 and 3 vs 3 no differences were found. Insight into this problem has direct consequences for sports vision, sports psychology, coaching and other training areas in sport. If a rapid movement of the eyes is needed by team players in order to detect the fast action within their sport, then this requires the training of dynamic ballistic eye movements, accompanied by an ability to quickly detect and process critical cues. But if an economical movement of the eyes defines how elite athlete read the playing environment, then a completely different approach is needed requiring the anticipation of critical events, the use of fewer fixations of longer duration and a more parsimonious but knowledgeable processing of relevant task information. Gaze of Athletes In Open and Closed Sports In order to explore this problem, the gaze behaviours of Team Canada ice hockey players were recorded on ice as they performed defensive plays standard in ice hockey using the vision-in-action (VIA) paradigm (Vickers (1996). The VIA paradigm has been adopted by a number of researchers to determine the gaze behaviors of athletes performing skills in live sport settings (Frehlich, 1997; Frehlich, Singer & Williams, 1999; Harle & Vickers, 2001; Janelle, Hillman, Hatfield, 2000; Vickers, 1996; Vickers & Adolphe, 1997; Williams, Singer, & Frehlich, in press). The VIA system (Vickers, 1996) simultaneously measures the subject s gaze, ocular, and motor behaviors while they perform selected skills or tactics in their sport. Using this approach in closed skills such as basketball (Vickers, 1996; Harle & Vickers, 2001), billiards (Frehlich, 1997; Frehlich, Singer & Williams, 1999; Williams, Singer & Frehlich, in press), darts (Vickers, Rodrigues & Edworthy, 2000) and shooting (Janelle et al, 2000), the gaze of elite performers has been found to exhibit a longer duration and lower frequency of fixation to specific objects or events. This fixation, which has been called a quiet eye is also located just prior to the initiation of the final aiming movement, therefore it has a consistent temporal location across sports. While a consensus across studies exists in closed sports in open sports such as volleyball (Vickers & Adolphe, 1997), table tennis (Rodrigues, Vickers & Williams, in press) and soccer (Williams et al, 1994, Williams, et al, 1998) the results are not clear. Some studies reporting no difference in fixation duration due to expertise (Rodrigues et al, in press), long and short durations of fixation within the same sport (Williams et al, 1994, Williams, et al, 1998). It has also been difficult to determine the temperal order of critical fixations in open sports, as compared to closed. Do experts see critical information earlier than non-experts or later? Together these findings show that defining the gaze behaviour of athletes in tactical settings presents a considerable challenge to researchers.

The goal of this pilot study was therefore to try and resolve some of the debate by analyzing the fixation/tracking behaviours of elite ice hockey players as they performed defensive plays in ice hockey. Fixation/tracking (F/T) behaviours occurred when the gaze was stable on a location, object, or person within the tactical setting. Of specific interest was the determination of the frequency, temporal sequence, duration, and location of F/T. Differences in F/T due to skill level and trial outcome (plus-minus) were also of interest. Method Participants Two Team Canada male ice hockey players, aged 20 and 24 years, were selected by the coaches of Team Canada based on their proficiency in playing defense at the international level in ice hockey. The expert (E) player was known for his exceptional defensive play, while the near-expert (NE) consistently made errors in decision-making while playing defense. The selection of the two players was confirmed by each players plus-minus statistic in a recent international tournament. The plus-minus score is a standard measure used in ice hockey that is calculated by subtracting the number of goals scored when the player is on the ice, from the number of goals scored by the opponent. Ice hockey teams are typically organized into 3 or 4 lines, with three forwards and two defensive players on each line. Each line typically plays for a 1-2 minutes and then is replaced in a constant rotation that goes on throughout the game. If a line is constantly scored against by the opponent then all members are negatively affected in terms of the plusminus statistic. The E player had a plus-minus score of + 6 in the tournament, meaning his team scored 6 more goals than the opponents when he was on the ice, while the NE had a plus-minus score of 5, indicating a major deficiency in preventing the other team from scoring. Figure 1. A frame of Vision-In-Action data as collected on ice in ice hockey. The eye image (A) shows the eye of the athlete with horizontal and vertical axes at pupil and corneal reflection centroids. The gaze image (B) shows the field of view in front of the athlete and location of the gaze in the hockey environment as indicated by the white cursor. The motor image (C) shows the actions of all players as recorded by an external camera. Time is shown in frames and was recorded at 30 frames per second, or 33.33 ms per frame.

Apparatus The Vision-in-Action (VIA) system (Vickers, 1996) was used on ice to record the gaze, ocular and motor behavior of the athletes. Applied Sciences Laboratories (ASL) 501 eye tracker was interfaced to an external video camera (Sony, Model TRV82) and two digital video mixers (Videonics, Model MX-1) to produce frames of video data similar to that shown in Figure 1. The ASL 501 is a monocular corneal reflection system that measures eye-line-of-gaze with respect to the helmet. The helmet has a 30-metre cord attached to the waist, connected to the eye control unit, thus permitting the athletes near-normal mobility. Miniaturized optics (scene and eye cameras), an illuminator, and visor were mounted on the helmet, with a total weight of 700 g. The VIA system has an accuracy of 1 degree of visual angle, precision of.5 degree, and an output frequency of 30 Hz. Protocol Data was collected on the Olympic Oval ice hockey rink at the University of Calgary. In order to prevent anticipation of set plays, three different defensive plays were presented in random order that were familiar to the players, therefore the participant did not know which tactical problem would be defended against. Each trial was skated against teammates, also on Team Canada, who assumed the role of defensive teammates or offensive opponents. Five trials were skated per player of each defensive play (15 total) to avoid problems of fatigue and familiarity with the plays. Each play included three temporal phases: pattern recognition (PR), situation assessment (SA) and decision-making (DM), which were derived from Klein (1998). The contain play is shown in Figures 2. At the outset of the PR phase, the defensive participant (DT) began each trial with eyes closed at center-ice in a position to respond to the movements of two defensive team-mates (e.g. defenseman (D1), defensive forward (Dt) and two offensive opponents (e.g. first offensive forward (O1), second offensive forward (O2) who skated through the neutral zone toward him. On the first whistle, the coach (B in Figure 2) shot the puck into the defensive zone at the same time that D1, DT, O1, and O2 skated toward Dp at game intensity and speed. Once (D1) passed a pre-determined marker on the ice, a second whistle was blown by the coach signaling Dp to open his eyes, identify and defend against the play. The play ended with a third whistle which signaled the end of the DM phase. Figure 2. The recognition (PR), situation assessment (SA) and decision-making (DM) phases of the contain play. The defensive participant (Dp) and cable holder (Ch) are shown as well as defensive player one (D1), defensive teammate (Dt), offensive player one (O1), and offensive player two (O2). The location of the VIA system (A), coach (B) and path of the puck (dotted line) at the beginning of each trial is also shown.

Each phase was defined by a definitive skating action that signaled the transition from the PR to the SA phase, and SA to DM phase. The PR phase onset occurred with the first appearance of Dp s gaze and the offset when he took the first skating movement to counter the movements of O1 and O2. Onset of the SA phase began with the first frame following the onset of skating movement of Dp. The onset of the DM phase began with Dp final skating movement to counter the play and offset with a third whistle signal. The approximate location on the ice of the PR, SA and DM phases is shown by the shaded portions of Figure 2. Coding and Analysis of Data The plus-minus quality of each play was determined by the coaches of Team Canada. The contain play was selected for in-depth analysis as two plays for the E were ranked as plus for the E player and two as minus the NE. The VIA data of the contain play was coded in an editing suite equipped with frame by frame shuttle control using coding rules adapted to ice hockey from previous studies (Vickers, 1996; Vickers & Adolphe, 1997; Williams et al, 1999). Due to the dynamic nature of ice hockey, no attempt was made to distinguish between fixation and tracking gaze. F/ T gaze were defined using minimum duration parameters from the eye movement literature (Carpenter, 1988) and occurred when the gaze was stable on a location, object, player or players for a minimum duration of 99.9 ms (3 or more frames). Frequency of F/T was determined, as was relative duration (%) and temporal order within each play (F/ T1, F/T2...F/Tn). F/T gaze locations were identified as offensive-defensive combinations (O-D combo), offence, defence, open space, puck, and other. Other included blinks, gaze locations beyond the task area, and gaze that were not codeable due to excessive head movement causing blurring of the gaze data. Since the duration of each trial and phase varied, absolute measures of F/T (ms), phase duration (ms) and gaze duration (ms) were converted to relative time (%) in a normalization procedure which expressed each variable as a percent of the phase duration. Thus, every trial had its onset transformed to (0%), its offset to 100%, and each point in between a proportion of the total time (Schmidt & Lee, 1999). Figure 3. Mean duration (in milliseconds) of the pattern recognition, situation assessment and decision making phases in ice hockey defensive play.

Figure 4. Temporal sequence and relative duration (%) of fixation/tracking (F/T) in the situation phase (SA) for the expert (E) athlete on plus plays and the near-expert (NE) on minus plays. F/T 9 occurred at the onset of the SA phase an,.f/t1 at the offset. Figure 5. Location and mean relative duration of fixate/tracking (F/T) of the expert and near-expert hockey players in the SA phase. Relative duration of F/T was calculated as a percent of the total SA phase duration.

Results and Discussion Duration of Phases The mean duration (ms) of each phase (PR, SA, DM) is shown in Figure 3. The PR phase averaged 722.95 ms (SE = 33.47 ms ), the SA phase 5204.22 ms (SE = 54 ms), and DM 1562.45 ms (SE = 603 ms). In the PR phase, both E and NE determined the defensive situation faced in a very short duration prior to initiating the first skating movement. The SA phase extended half the length of the ice (see Figure 2) and lasted for a long ~ five seconds. During the SA phase the athletes scanned the play and assumed their assigned roles, which involved countering the movement of the offensive players and co-coordinating their actions with defensive teammates. The DM phase averaged a second and a half. During this time the final cues were detected and movements made to counter the play leading to the end of the trial. Frequency and Temporal Sequence of F/T No differences in F/T were found in the PR phase, with both players averaging one F/T.yers. No results are presented for the DM phase due to F/T in the SA phase extending into the final phase of the play. These F/T are therefore analyzed as part of the SA phase. Figure 4 presents the mean relative percent of F/T in the SA phase for E on plus plays and NE on minus plays. The temporal sequence of F/T is shown with F/T 9 occurring at the onset of the phase and F/T 1 at the offset. The NE player recorded a higher frequency of gaze than the E player, nine for the NE and six for the NE. Figure 4 also shows that F/T durations were shorter at the beginning of each trial than at the end for both, E and NE. Figure 5 presents the location and mean duration of F/T across the plus trials for E and minus for NE. The E player allocated more time to each location than the E, ranging from a low of 7% of total phase duration to a high of 20%. The E player looked for the longest duration at the 0-D combination, followed by space, and least to defensive teammates. The NE player s F/T were similar to the E in that the same locations were fixated but for shorter durations. It is interesting to note that neither player fixated the puck during any of the trials, indicating the ambient system must have been sufficient to keep track of the movement of the puck. Conclusion These results explain, in part, why research in vision in tactical sports has been equivocal. Recall that Helsen and Pauwels (1993) found a lower search rate and longer duration of fixation in a soccer shooting situation on goal, while Williams et al. (1994) and Williams et al (1998) found different search rate if the players were playing in a 1 v1, 3 vs 3 and 11 on 11 situations. Williams et al (1998) suggested the reason for the different search characteristics lie in spatial qualities of each play. Our results support this argument. When the tactical problem was distant from the players (see Figure 2) and required the scanning of a wider field of view, then a higher frequency and lower duration of F/T was found (see Figure 4). When the tactical situation moved closer to the athlete the perceptual information acquired was dictated by a decrease in distance between the participants and other players and the aspects of the scene that could be seen by the participant. During the latter part of defensive tactical

play, defensive roles also become more narrowly defined and specific. Since the data was collected in a live on-ice setting, then each trial began with a wide spatial area, which narrowed and became more specific as the trial duration progressed. A lower duration of F/T was found for both E and NE at the beginning of each trial consistent with the need to evaluate an extensive number of perceptual information sources. The field of play was also wider and at a greater distance from the athletes. The shorter duration of F/ T during the first part of each trial was due to greater space within which the opposing players and teammates were located thereby requiring a faster shift of gaze at the beginning of the trial and shorter duration of fixation than at the end. During the latter part of the trial, the athlete s role within the tactical problem required they be responsible for 1on 1 assignments, and in this situation a lower frequency of high duration of F/T was found due to a decrease in the perceptual information sources. Recommendations Some caution should be used interpreting these results due to only two athletes being represented. Within this limitation, a few recommendations can be made. Earlier it was mentioned that if the rapid movement of the gaze is needed to detect the fast action typical of team sports, then this would require a different approach to training the visual, perceptual and cognitive systems than if a more economical control of the gaze defines expertise. To be effective, vision trainers, sport psychologists, coaches and others involved in training athletes need to develop both far and near fixation/tracking training programs for tactical skills, with the guiding principal being the position of the athlete relative to the playing surface, the complexity of the tactical play, and the distance of the action from the player. When the tactical problem is distant and includes a large number of opponents and teammates, the athlete needs to be able to detect a greater number of critical cues using a higher frequency of fixation/tracking of low relative duration. As the play nears, he/she should narrow their focus to fewer locations with a lower frequency and longer duration of fixation/tracking due to the proximal nature of the play and immediacy of the decisions and actions required. Perceptual program specific to sports have been developed by Wiliams et al, that can be adapted to both setttings (Williams, Ward + Champman submitted). References Abernethy, B. (1990). Expertise, visual search, and information pick-up in squash. Perception, 19, 63-77. Abernethy, B. & Russell, D. G. (1987). Expert-novice differences in an applied selective attention task. Journal of Sport Psychology, 9, 326-345. Carpenter R. H. S. (1988). Movements of the eyes, London: Pilon. Frehlich, S. G., Singer, R. N., & Williams, A. M. (1999). Visual search pattern in experienced and inexperienced billiards. [Abstract]. Journal of Exercise and Sport Psychology: Supplement. p. S38. Harle, S. & Vickers, J.N. (2001). Training quiet eye (QE) to improve shooting accuracy in the basketball free throw. The Sport Psychologist, 15, 289-305. Helsen, W. & Pauwels, J. M. (1992). A cognitive approach to visual search in sport. In D. Brogan & K. Carr (Eds.), Visual Search, Vol. II, London: Taylor and Francis.

Klein, G. (1998). Sources of power: How people make decisions. Cambridge Mass: The MIT Press. Janelle, C., Hillman, C., & Hatfield, B. (2000). Concurrent measurement of electroencephalographic and ocular indices of attention during rifle shooting: An exploratory case study. International Journal of Sports vision, 6 (1), 21-29. Ripoll, H. (1991). The understanding-acting process in sport: the relationship between the semantic and the sensorimotor visual function. International Journal of Sport Psychology, 22, 221-243. Rodrigues, S. T., Vickers, J. N., & Williams, A. M. (in press). Head, eye, and arm coordination in table tennis: An exploratory study. Journal of Sport Sciences. Schmidt, R. A. & Lee, T. D. (1999). Motor Learning and Performance. Champaign. Il. Human Kinetics. Vickers, J. N. (1996). Visual control when aiming at a far target. Journal of Experimental Psychology: Human Perception and Performance, 22 (2), 342-354. Vickers, J. N. & Adolphe, R. M. (1997). Gaze behaviour during a ball tracking and aiming skill. International Journal of Sports Vision, 4 (1), 18-27. Vickers, J. N., Rodrigues, S. T., & Edworthy, G. (2000). Quiet eye and accuracy in dart throw. International Journal of Sport Vision, 3 (1), 54-60. Williams, A. M. & Davids, K. (1994). Eye movements and visual perception in sport. Coaching Focus, 26, 6-9. Williams, A.M., Ward, P. and Champan, C. (sumitted) Research Quarterly Exercise and Sport. Williams, A. M. & Davids, K. (1998). Visual search strategy, selective attention and expertise in soccer. Research Quarterly for Exercise and Sport, 69 (2), 111-129. Williams, A. M., Davids, K. & Burwitz, L. (1994). Ecological validity and visual search research in sport. Journal of Sport and Exercise Psychology, S16-22. Williams, A. M., Davids, K. & Williams, J. G. (1999). Visual Perception and Action in Sport. London: E & FN Spon. Williams, A. M. & Grant, A. (1999). Training perceptual skill in sport. International Journal of Sport Psychology, 30, 194-220. Williams, A. M., Singer, R.A. & Frehlich, S. (in press). Quiet eye duration, expertise and task complexity in a near and far aiming task. Journal of Motor Behavior.