Visual cues and information used to anticipate tennis ball shot and placement.

Shim, Jaeho ; Miller, Glenn ; Lutz, Rafer 等

In sports where motor responses produced by athletes have inherent time constraints, the types of visual information athletes perceive or use may have a strong influence on their performance. Successfully returning a tennis serve or blocking a penalty kick in soccer is likely related to an athlete's ability to use less time processing, or better exploiting, visual information in his or her environment. Different sports make different demands on an athlete's perceptual systems, and successful performance requires that an athlete attend to important information, and ignore irrelevant information, in the sport environment. In tennis, for example, one must interpret how the opponent's movements relate to different types of volleys or shot returns. It is important to determine how expert and novice athletes differ in their use of visual information so that instruction may be improved. In addition, the investigation of visual cues and their use in sport has the promise to help scientists better understand movement control and expertise.

The study of perception in sports has often used one of two protocols--visual search pattern recording or visual occlusion techniques. For example, an eye-movement recorder has been used to determine the visual search pattern of sport performers and to track where, how long, and in what order their eyes fixate on different parts of the environment (Goulet, Bard, & Fleury, 1989; Singer, Cauraugh, Chen, Steinberg, & Frehlich, 1996; Ward, Williams, & Bennett, 2002; Williams & Davids, 1998; Williams, Davids, Burwitz, & Williams, 1994; Williams, Ward, Knowles, & Smeeton, 2002). In contrast, the visual occlusion technique masks an opponent's body parts (e.g., legs, trunk, and arm) and/or equipment (e.g., racquet and soccer ball) using film (Abernethy, 1988; Abernethy & Russell, 1987a, 1987b) or video editing (Williams & Davids, 1998).

One of the popular methods for measuring eye movements is head-mounted corneal-reflection. This method uses the reflection of a beam of light placed in front of the cornea to form an image which is video recorded. A change in the point of fixation changes the position of the cornea. Eye movement recorders have been used to investigate perception in many sport settings such as soccer (Williams & Davids, 1998; Williams, Davids, Burwitz, & Williams, 1994), gymnastics (Bard, Fleury, Carriere, & Halle, 1980), ice hockey (Bard & Fleury, 1981), tennis (Goulet, Bard, & Fleury, 1989; Petrakis, 1986; Singer et al., 1996), golf (Vickers, 1992), volleyball (Vickers & Adolphe, 1997), baseball (Bahill & LaRitz, 1984), table tennis (Ripoll & Fleurance, 1988), and basketball (Vickers, 1996). These studies have generally demonstrated that experts are able to use perceptual resources with greater efficiency. In general, experts demonstrate fewer fixation points on important environmental information but longer fixation durations. For example, Goulet et al. (1989) found that expert tennis players fixate their eyes more on the opponent's arm and racquet whereas novices fixate primarily on the ball. Singer et al. (1996), on the other hand, did not find large differences comparing experts' and novices' visual scanning profiles, except that novices tend to fixate more on the head of the opponent during the serve. However, experts were more effective than novices in anticipating type of serve, direction of serve, and direction of groundstroke.

Studies of expert's and novice's use of visual cues in sport environments have been mostly equivocal on what body parts are important focus points. While there are several reasons for the lack of conclusive results, one major limitation of studies using eye tracking equipment is that participants' visual orientation may not be directly related to the point where information is extracted and that attention can relocate within the visual field without an eye movement (Abernethy, 1988). For example, a guard dribbling down the basketball court may look at a distant point in the visual field down court, but attend to the periphery to find an open player. Athletes may actively seek information from the point where they fixate their eyes when returning a tennis serve, batting in baseball, and blocking a penalty kick in soccer. However, athletes may also fixate their eyes on a point remote from where they intend to throw, shoot, or hit to disguise their action or to deceive their opponents.

Because attention can shift without an eye movement, some caution must be taken when interpreting results obtained using an eye tracking equipment. The spatial occlusion technique may be a more viable method for investigating expert's perceptual skills. The spatial occlusion technique has been used to investigate perceptual skills of badminton (Abernethy and Russell 1987a and 1987b), squash (Abernethy 1990b), and soccer players (Williams & Davids, 1998). The spatial occlusion technique selectively occludes specific body parts or sport equipment for the duration of the trial. When using this technique, researchers are making the assumption that if the ability to anticipate movement outcome is reduced when a specific body part and/or equipment is occluded, then the occluded visual cue must be important. Abernethy (1990b) and Abernethy and Russell (1987a) occluded the racquet, racquet and arm, head, and lower body of an opponent player by coloring each film image with black ink in rectangular shape. Results of these studies showed that the occlusion of the racquet and arm cause the greatest anticipation error among the different occlusion conditions, indicating that this information is very important in racquet sports. In Williams and David's soccer study (1998), different body regions were occluded such as the head and shoulders, hips, and lower leg and ball region. The results showed that, despite the difference of expert and novice anticipation, occluding the opponent dribbler's hips, lower leg, or ball region does not affect expert anticipation more than novice anticipation, which suggests that experts are able to acquire similarly valuable information from non-occluded areas of display.

In all of the occlusion conditions in the racquet sports studies, though tests of statistical significance were not reported, players anticipated the ball destination at a level greater than chance. In Abernethy's study (1990b), the error percentages of both experts and novices in the occlusion conditions were less than chance error (50%). The error percentage for experts was less than 20% when anticipating lateral direction and less than 10% when anticipating depth. For novices, lateral error percentage was slightly over 40% and the depth error percentage was below 30%. These results indicate that there may be visual cues other than the racquet and arm that provide important information.

In past years, researchers have made strides toward making the lab settings that better approximate the real sports environment by moving from a small video screen (Singer et al., 1996) and verbal (Abernethy, 1990a, 1990b) report to a large screen (Williams et al., 2002) or live opponent (Abernethy, 1990a, Singer et al., 1996) and motor responding (Singer et al., 1998; Williams et al., 2002). It was the goal of the present investigation to build on these attempts to increase ecological validity by matching the testing conditions very closely to an actual tennis environment. To achieve these ends, players viewed a life-sized opponent on a screen located where the opponent would actually stand on the opposite baseline. In addition, players' responses consisted of actual simulated motion to the video display.

Despite the limitation that an eye movement recorder can determine only the eye fixation point and not the focus point of attention, it has been used extensively in the studies of perception in sport. However, the spatial occlusion technique has been used in only a few studies because of editing difficulties. With recent advances in video editing technique, the use of occlusion can be a valuable tool to study visual perception in sports. The purpose of this study was to determine what and how visual cues contribute in determining an opponent's stroke outcome. We used additional occlusion conditions from previous studies to examine this question more comprehensively. It was hypothesized that the racquet and forearm display would provide the most valuable information to players for anticipating ball outcomes, but that other body parts would provide important information for different types of ball outcome.

Method

Participants

Fourteen (7 males and 7 females) highly-skilled (national tennis rating of 5.5 or above) intercollegiate tennis players and fourteen (7 males and 7 females) novice college tennis players participated in this study. The rating system developed by the United States Tennis Association rates beginning players at 1.0 and professional players at 7.0. With a rating of 5.0, players are described as being able to make good shot anticipation. Another highly-skilled male player (5.5), whose recorded and edited tennis strokes were projected on a screen, served as the video opponent of each participant. All participants signed an informed consent form approved by the university institutional review board.

Apparatus

A S-VHS (Panasonic AG-456U) camera was used to record the strokes performed by the opponent and to record the player's anticipatory response. A Sharp LCD projector (Notevision 2SB, brightness = 1400 ansi lumens) was used to project the recorded image on a 2.7 m screen (Da-Lite).

Recording and Editing Procedure

The "opponent's" movements were recorded on an indoor tennis court with the camera positioned midway between the net and service line, opposite of the opponent's courtside. Four strokes (forehand down-the-line and cross-court, and topspin lob to forehand and back-hand corner) were videotaped and edited using digital computer graphics techniques (Adobe Premiere and Photoshop). First, the strokes were "digitized" and placed in video clips on the computer. Then, each frame was edited to occlude body parts of the opponent using dark green. These edited video clips were then replaced on videotape for use in the experiment. Occlusions were made so that only the following body parts and/or racquet of the opponent were displayed: (1) head, (2) racquet and forearm, (3) trunk, (4) lower body, and (5) opponent in full where no occlusion was done.

The five display conditions were produced for each of the four strokes, and these 20 edited sequences were duplicated three times and placed in random order on an S-VHS tape, resulting in 60 sequences.

Experimental Procedure

The participant stood in "the ready position" when viewing the opponent's shots at a location marked 3 m behind the net facing the screen on which the opponent's image was projected. The screen was positioned at the baseline where the opponent stood when the shots were recorded. The projector was placed on the floor near the service line on the opponent's courtside so that the height of the opponent projected on the video screen would match his actual height. The ball and the opponent were both displayed on the screen, and as the opponent made contact with the ball, the screen turned to green. At this point, the participant had to move towards the anticipated location of the opponent's shot and simulate hitting the appropriate return (forehand volley, backhand volley, forehand overhead from baseline corner, or backhand overhead from baseline corner) as quickly as possible. Although this response is slightly different from that required by the actual task because no ball is actually struck, it is much more realistic than a verbal or pen-and-paper response. In this manner, participants were able to use the motor system to respond to the visual cues they perceived.

An S-VHS camera (sampling at 60 Hz) was positioned facing the participant on the opposite (opponent's) courtside to record the movements of the participant. The camera captured the participant's racquet response and a small mirror that reflected the opponent's stroke on the screen (Figure 1). This set-up allowed a determination of the player's initial racquet response and instant of ball-racquet contact made by the opponent, and subsequent calculation of response delay time.

[FIGURE 1 OMITTED]

Data Reduction and Analysis

Percentages of correct shot anticipation were calculated for each of the five occlusion conditions. For example, if the player correctly anticipated the opponent's shots nine out of 12 times as it was presented in the racquet and forearm display condition, stroke anticipation accuracy for that display condition would be 75%. Response delay time was calculated as the time elapsed from the opponent's ball-racquet contact to the player's initial movement of the racquet to hit their simulated shot. The instant of racquet movement was operationally defined as the moment when the player's dark wrist band horizontally displaced more than 30 cm from the midline of the player's trunk. Oftentimes, the players initiated the racquet movement in one direction and quickly changed to the other direction. Therefore, a displacement of 30 cm was used to allow for the correction during anticipation. It should be noted that the response delay time was measured not for the purpose of measuring an exact reaction time but for the purpose of comparing between display conditions. Therefore, we consider the procedure acceptable for the latter purpose. Two raters simultaneously observed the video of each player and the instant of racquet movement was determined when both raters agreed.

Statistical analyses were performed on participants' stroke, stroke direction (down-the-line and cross-court), and stroke type (groundstroke and lob) anticipation accuracy and response delay time. Inspection of anticipation percentages revealed a non-normal distribution of scores. Thus, scores were transformed to arcsine values to satisfy requirements for normality. To compare stroke anticipation accuracy and response delay time, 2 x 5 (Skill x Display) ANOVAs with repeated measures for display were conducted. For any violation of sphericity, Greenhouse-Geisser procedure (Schutz & Gessaroli, 1987) was performed to adjust the degrees of freedom. For all significant main effects, follow-up multiple comparisons (Bonferroni) were performed. In addition, one-sample t-tests were performed on anticipation accuracy for each display condition to determine if anticipation accuracy was greater than that of chance occurrence.

Results

For the repeated measures ANOVA on stroke anticipation accuracy, Mauchley's sphericity test ([P.sup.2] = 19.7, p < .05) showed a violation of the assumption of sphericity, and therefore a Greenhouse-Geisser procedure was used to adjust degrees of freedom. There were significant effects for skill, F(1, 26) = 14.2, [beta] = .95, p < .01, and display, F(2.77, 72.06) = 4.3, [beta] = .95, p < .01, but there was no significant interaction between skill and display (p > .05). Experts showed a greater accuracy in stoke anticipation than novices, and post-hoe (Bonferonni) analyses showed significantly greater anticipation accuracy in the normal display compared to the head display condition (p < .05). Also, one sample t-tests showed that players anticipated shots at a level greater than chance in all display conditions (ps < .05), except in the head display condition (p > .05). Descriptive statistics for anticipation accuracy and response delay time are summarized in Table 1 and Figure 2.

[FIGURE 2 OMITTED]

Repeated measures ANOVA on stroke direction anticipation accuracy yielded no significance for any main effects or interaction (Fs < 1). When ANOVA was performed on the stroke type (groundstroke or lob) anticipation accuracy, significant effects were found for skill, F(1, 26) = 26.8, [beta] = .99; display, F(4, 104) = 10.2, [beta] = .99; and skill and display interaction, F(4, 104) = 3.6, [beta] = .86, ps < .01. Expert players were able to anticipate stroke type with less error than novice players. Post-hoe (Bonferonni) analyses showed that the normal (no occlusion) display resulted in significantly greater stroke type accuracy compared to the head, trunk, and legs displays (ps < .01) with the exception of the racquet and arm display (p >.05). To determine which two display conditions showed significant interactions with skill, 2 x 2 (Skill x Display) ANOVAs were performed with adjusted alpha level (0.05/10 = 0.005) due to possible alpha inflation caused by multiple ANOVAs tested. The racquet and legs displays showed significant interaction with skill, F(1, 26) = 9.50,p < .005, and so did the legs and normal displays with skill, F(1, 26) = 9.50,p < .005 (Figure 3).

[FIGURE 3 OMITTED]

The ANOVA on response delay time showed a significant effect for display, F(4, 104) = 6.6, [beta] = .99, p < .01, while no other main effect or interaction reached significance (ps > .05). A post-hoe (Bonferonni) analysis showed significantly faster response delay time in the normal display condition compared to the head display and trunk display conditions, ps < .01 (see Figure 2). Interestingly, experts did not respond to correctly anticipated shots any faster than the falsely anticipated shots, t(1, 13) = 1.1,p > .05.

In summary, expert players showed less error for anticipating ball outcomes than novices, but they did not show faster response delay times. Player anticipation was poorest and response delay time was longest in the head display condition, while anticipation accuracy was highest and response delay was shortest in the normal display condition. Excluding the normal display condition, players improved stroke type anticipation more in the racquet and forearm display than in other display conditions.

Discussion

All body pans and the racquet, with an exception of the head, included in this study appear to provide some information to participants that allow for more effective anticipation or faster reaction to the video display of an opponent. The results showed that not only the racquet and forearm but also the trunk and legs display conditions allow players to anticipate tennis shots at a level greater than chance. Generally, players had a greater ability to anticipate stroke type (depth) compared to stroke direction. In Abernethy and Russell's study (1987a & 1987b) on badminton, players produced a greater lateral error in predicting the shuttle cock landing position compared to depth error. Similar results were also found in squash (Abernethy, 1990b) which corroborate the findings of this study.

Among the body parts tested, the motion of the racquet and forearm appeared to provide the most valuable visual information to the players. However, the racquet and forearm seem to be particularly important for determining stroke type because, for stroke direction, no effect was found for skill level or display condition. Similarly, Abernethy (1990b) found the racquet and arm motion to be a more important visual cue for determining depth than for directional outcome of the opponent's swing in squash. The racquet and forearm display condition includes information about ball contact time and position based on the traveling speed of the racquet and ball. Ball contact time and position are important features for determining directional ball outcome, but that information did not enhance the ball direction anticipation greater compared to other displays. In contrast to Abernethy's (1990b) finding in squash, Abernethy and Russell's (1987a) finding in badminton showed rather that the racquet and arm was more important for stroke direction than stroke depth. Because of the light weight of a badminton racquet, stroke direction can be more easily manipulated through a flick of the wrist whereas the wrist movement may be more limited in tennis and squash due to the racquet weight. Therefore in badminton, observing the wrist action through the racquet and forearm may be important in determining stroke directions.

Besides the racquet and arm, other body segments appear to provide important visual information. The trunk and lower body appear to provide some information on the directional outcome of ball strike. A greater shoulder turn and closed stance are intended more for down-the-line shots whereas a moderate shoulder turn and open stance are intended more for crosscourt shots. However, these movement characteristics are now difficult to see, or at least slowly diminishing in modern tennis technique. With recent changes in the striking pattern and players becoming more skillful in disguising their shots, the racquet and arm movement may be the only salient visual cue in determining shots. While various body parts have more room for faking and disguising movements because they do not directly affect the ball outcome, the racquet movement near impact cannot be faked because the racquet orientation at impact and pre- and post-impact racquet velocities directly affect the ball outcome. Traditionally in tennis, the down-the-line shots are instructed to be performed with a closed stance and cross-court shots with a more open stance while transferring a great amount of weight to the leading leg. Now, players are taught to limit their weight transfer but are encouraged to produce a greater shoulder and hip turn and to place more weight on the rear leg as in baseball batting. An EMG study (Knudson & Blackwell, 2000) of the trunk muscles showed no significant differences in muscle activation between the open and square stance when hitting forehand drives, and a kinematics study (Knudson & Bahamonde, 1999) of the trunk and racquet showed no significant differences of racquet resultant velocity, vertical path of racquet, and trunk angular velocity at impact between the two drives. With little differences found between the two stances, players and coaches may prefer the open stance because it makes it easier to disguise shots and allows for a better body transition after striking the ball. An open stance also promotes a greater body turn.

The racquet and arm movement can provide diverse information. To hit a drop shot in badminton and squash, the player quickly decelerates the racquet and arm swing before impact. To hit a topspin lob in tennis, the player drops the racquet and arm considerably lower than when hitting a groundstroke. The racquet and arm swing is also slower when hitting a topspin lob than hitting a groundstroke. Striking a ball down-the-line or cross-court can also be accomplished different ways. The ball can be stroked in two directions by changing only the ball-racquet contact points while maintaining the same swing. Contacting the ball early in the forward swing would lead to a down-the-line shot whereas contacting the ball later in the end of swing would lead to a cross-court shot. Instead of the ball-racquet contact location, the racquet position at impact can be changed to hit in different directions. The wrist can be cocked through impact so that the racquet head is facing down-the-line direction or the wrist can be flexed through impact to hit a cross-court shot.

In an actual tennis match, players appear to anticipate infrequently and rarely do they anticipate incorrectly. Instead, players usually wait until they have sufficient information about the ball direction before moving. In most cases, players wait until they observe the initial ball flight. Therefore, skilled players have the advantage of responding more quickly to the ball and not necessarily the advantage of making a correct response selection. A previous study (Shim, Carlton, Chow, & Chae, 2005) has shown that skilled tennis players had shorter response latency when returning a groundstroke performed by a live hitter compared to a ball machine. The difference in the response latency was dramatic. Players were more than 25% faster when they could see the movement pattern of the hitter. The 50 ms time saving and means that skilled players have an additional 50 ms to move, and this would allow the player to increase his/her court coverage by as much as 1.2 m (0.6 m on both the forehand and backhand side).

Players are now becoming more adept at disguising their shots. They try to maintain same swing until the racquet reaches close to impact. With the movements of all other body parts consistently controlled, the movement pattern of racquet and arm, which will eventually determine the ball outcome, will become more important. The opponent that players viewed in this study was asked to produce normal shots which may be different from the shots produced in a match. A future study should focus more on the racquet and arm with variations on their movements and their effect on anticipation.

Table 1
Anticipation Accuracy and Response Delay Time as a Function of Skill
Level

 Skill level

 Expert Novice

Stroke 41 32
 Anticipation (07) (06)
 accuracy (%)

Stroke 60 58
 Direction (07) (16)
 Anticipation
 accuracy (%)

Stroke Type 67 56
 Anticipation (06) (15)
 accuracy (%)

Response 350 420
 delay (90) (180)
 time (ms)

Note. Values in parentheses indicate between-subject standard
deviations.

References

Abernethy, B. (1988). The effects of age and expertise upon perceptual skill development in a racquet sport. Research Quarterly for Exercise and Sport. 59(3), 210-221.

Abernethy, B. (1990a). Expertise, visual search, and information pick-up in squash. Percep tion, 19, 63-77.

Abernethy, B. (1990b). Anticipation in squash: Differences in advance cue utilization between expert and novice players. Journal of Sports Sciences, 8, 17-34.

Abernethy, B., & Russell, D. G. (1987a). Expert-novice differences in an applied selective attention task. Journal of Sport Psychology, 9, 326-345.

Abernethy, B., & Russell, D. G. (1987b). The relationship between expertise and visual search strategy in racquet sport. Human Movement Science, 6, 283-319.

Bahill, A., & LaRitz, T. (1984). Why can't batters keep their eyes on the ball? American Scientist, 72, 249-253.

Bard, C., & Fleury, M. (1981). Considering eye movement as a predictor of attainment. In I. M. Cockerill & W. W. MacGillvary (Eds.) Vision and Sport (pp. 28-41). Cheltenham: Stanley Thournes.

Bard, C., Fleury, M., Carriere, L., & Halle, M. (1980). Analysis of gymnastics judges visual search. Research Quarterly for exercise and Sport, 51, 267-273.

Goulet, C., Bard, C., & Fleury, M. (1989). Expertise differences in preparing to return a tennis serve: A visual information processing approach. Journal of Sport and Exercise Psychology, 11, 382-398.

Knudson, D., & Bahamonde, R. (1999). Trunk and racket kinematics at impact in the open and square stance tennis forehand. Biology of sport, 16(1), 3-10.

Knudson, D., & Blackwell, J. (2000). Trunk muscle activation in open stance and square stance tennis forehands. International Journal of Sports Medicine, 21(5), 321-324

Petrakis, E. (1986). Visual observation patterns of tennis teachers. Research Quarterly for Exercise and Sport, 57(3), 254-259.

Ripoll, H. & Fleurance, P. (1988). What does keeping one's eye on the ball mean? Ergonomics, 31(11), 1647-1654.

Schutz, R. A., & Gessaroli, M. E. (1987). The analysis of repeated measures designs involving multiple dependent variables. Research Quarterly for Exercise and Sport, 58(2), 132-149.

Shim, J., Carlton, L. G., Chow, J., & Chae, W. (2005). The use of anticipatory visual cues for highly skilled tennis players. Journal of Motor Behavior, 37, 164-175.

Singer, R. N., Cauraugh, J. H., Chen, D., Steinberg, G. M., & Frehlich, S. G. (1996). Visual search, anticipation, and reactive comparisons between highly skilled and beginning tennis players. Journal of Applied Sport Psychology, 8, 9-26.

Vickers, J. N. (1992). Gaze control in putting. Perception, 21, 117-132.

Vickers, J. N. (1996). Control of visual attention during the basketball free throw. American Journal of Sports Medicine, 24(6), S93-S97.

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.

Ward, P., Williams, A. M., & Bennett, S. J. (2002). Visual search and biological motion percep tion in tennis. Research Quarterly for Exercise and Sport, 73, 107-112.

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., & Williams, J. G. (1994). Visual search strategies in experienced and inexperienced soccer players. Research Quarterly for Exercise and Sport, 65, 127-135.

Williams, A. M., Ward, P., Knowles, J. M., & Smeeton, N. J. (2002). Anticipation skill in a real-world task: Measurement, training, and transfer in tennis. Journal of Experimental Psychology: Applied, 8(4), 259-270.

Author Note

This research was in part supported by a Baylor University Research Council Grant. The authors wish to thank Baylor University men's tennis coach Matt Knoll and women's tennis coach David Luedtke for their support of this research.

Jaeho Shim, Glenn Miller and Rafer Lutz

Baylor University

Address Correspondence To: Jaeho Shim, Baylor University, Department of Health, Human Performance, & Recreation, P.O. Box 97313, Waco, TX 76798 Phone: (254) 710-4009 Fax: (254) 710-3527 E-mail : [email protected]