Gaze control ability of League of Legends players in various game situations: Perspectives from solo-ranked match Inhyeok Jeong1, Donghyun Kim2, Naotsugu Kaneko1 and Kimitaka Nakazawa1 1 The University of Tokyo, 3-8-1, Komaba, Meguro-ku, 153-8902 Tokyo, Japan 2 MaxForm Corp., 242, Gonghang-daero, Gangseo-gu, 07805 Seoul, Republic of Korea Abstract Previous research analyzes the superior gaze control ability of esports players in simple cognitive tasks or in full games. Therefore, it is necessary to understand the gaze control ability of esports players in various situations. We assumed that game situations that require multiple tasks had wider gaze distribution than other situations. Therefore, the current study aims to compare the gaze control ability between high- and middle-skilled "League of Legends (LoL)" players and among various game situations classified into four categories and in an unclassified situation (five in total). Eight high-skilled (top 10%) and eight middle-skilled (lower than the top 10%) LoL players were recruited for the experiment. They wore an eye tracker and were asked to play solo-rank matches in LoL games. We analyzed gaze distribution, Region of Interest (ROI), and fixation duration during the games. The results showed that high-skilled players had a wider gaze distribution and shorter fixation time regardless of the game scene than middle-skilled players. Furthermore, high-skilled players checked the ROI area more frequently than middle-skilled players, where they could see the overall flow and feedback of the game. Thus, focusing on the overall flow and feedback with wide gaze distribution is the source of high performance in LoL players. When the game situations required focusing on multiple stimulations simultaneously, wide gaze distribution was observed rather than in other situations, regardless of the skill level. Our results suggest that it is necessary to adopt the appropriate gaze control training for esports players based on the various situations in esports. Keywords Gaze control ability, League of Legends,1solo-rank game, esports 1. Introduction with complex strategies [9]. The MOBA game generates various situations, such as one-on-one matches, team fighting, and communication with other players. Esports consists of competitive video games with League of Legends (LoL) is one of the most famous online and offline spectators [1]. With the esports belonging to the MOBA game, where players team development of the esports industry, research on up with 5 teammates to fight against opponents. To esports has also increased in various fields. From the achieve high performance in LoL, wide gaze distribution point of view of cognitive science, it is known that and short fixation time are important for collecting more esports can help players gain faster reaction time and information during gameplay [10]. Moreover, in real-time information processing skills [2, 3]. Moreover, esports strategy (RTS) games with a similar gaming interface to experts have superior visual behavior (e.g., gaze MOBA games, high-skilled RTS players had wider gaze movement) and attention skills (e.g., visual attention) [4, distribution with fast gaze movement than low-skilled 5]. The gaze movement is a well-known factor for RTS game players [4]. Information and interfaces in understanding the superior performance level in MOBA and RTS games are widely distributed across the esports [6]. Among the many esports genres, the entire monitor. Therefore, fast and wide gaze movement multiplayer online battle arena (MOBA) game is well allows high-skilled RTS and MOBA game players to known to require a high level of gaze control ability collect more information faster than low-skilled game and cognitive functions [7,8]. In MOBA games, players players during the gameplay [4,10]. To sum up the team up with other teammates to fight against opponents previous studies [4,10], it is uncontroversial that high- 8th International GamiFIN Conference 2024 (GamiFIN 2024), April 2- 5, 2024, Ruka, Finland hofwillson@gmail.com (A1, Inhyeok Jeong); devorca18@gmail.com (A2, Donghyun Kim); kaneko@idaten.c.u- tokyo.ac.jp (A3, Naotsugu Kaneko); nakazawa@idaten.c.u-tokyo.ac.jp (A4, Kimitaka Nakazawa) 0000-0002-1343-7442 (A1, Inhyeok Jeong); 0009-0001-9710- 2542 (A2, Donghyun Kim); 0000-0002-1587-9287 (A3, Naotsugu Kaneko); 0000-0001-5483-8659 (A4, Kimitaka Nakazawa) © 2024 Copyright for this paper by its authors. The use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). CEUR Workshop Proceedings (CEUR-WS.org) CEUR ceur-ws.org Workshop ISSN 1613-0073 Proceedings 1 skilled MOBA and RTS game players have superior Table 1 gaze control abilities. Specific information about participants However, LoL game players do not always have to pay Age High-skilled: 22.7 ± 3.0 attention to all the information from the entire monitor. In Middle-skilled: 22.7 ± 2.6 a game situation when players are engaged with multiple Education level High school: 13 opponents, it is important to focus on a single piece of University: 3 information. When strategizing the fight, it is important to pay attention to various information for an effective battle Sex Male: 16 simultaneously. Each percentage of the situation was LoL rank Bronze tier: 2 dynamically changed throughout the game. To sum up, Silver tier: 2 situations that need to focus on multiple information and Gold tier: 4 single information exist at the same time in a single LoL Platinum tier: 3 game match. Emerald tier: 1 Even though various game situations exist in LoL, the Diamond tier: 4 criteria for dividing the game scene for scientific research Experience High-skilled: 9.2 ± 1.4 in LoL is still lacking. Furthermore, given that different year Middle-skilled: 4.3 ± 4.0 game situations exist in a single LoL game match, it is necessary to reveal the situation-based gaze control abilities of LoL players. In RTS games, gaze movement 2.2. Equipment training based on the game situation has been suggested [4]. Therefore, revealing the situation-based gaze ability The current study used a 27-inch 144 Hz refresh rate of skilled LoL players, one of the most famous MOBA monitor for the experiment (ASUSTek Computer Inc., genre esports might contribute to developing a new Taiwan) and an eye tracker (Pupil-core, Pupil-Lab, specific training method for LoL players. Haftungsbeschränkt, Berlin, Germany). Gaze In the current study, we used solo rank games to movement was recorded using Pupil-Core, open-source categorize the game scene and evaluate the gaze control software for Pupil-Core (Version 3.5.7). The eye tracker ability of LoL players. The rankings of participants had one video camera (60 Hz, 1920 x 1080 px) and two directly change based on wins and losses matches in LoL eye cameras (200 Hz, 192 x 192 px) to record the solo rank games. This ranking system serves as an experimental environment and gaze movement, important motivational factor for esports players [11,12]. respectively. The eye tracker had 0.06° accuracy with Thus, requiring participants to play a solo-ranked game is calibration and 0.02° precision. The eye tracker used the an effective approach to studying their ability in actual “dark pupil” detection method to analyze gaze movement. game situations. To sum up, using the solo rank game is The “dark pupil” detection method detects the edge of the suitable for investigating the gaze control ability of LoL pupil for estimating the location of the gaze position. The players in various situations of actual LoL matches. eye tracker (pupil-core) had a maximum 40-millisecond The purpose of the current study is to investigate delay in processing the gaze movement data (including the gaze control ability of skilled LoL players in pupil image transport, formatting, detecting, and showing challenging and motivating tasks adapted to various the results) [13]. Four surface markers were attached to game situations using a solo rank game of LoL. The the corners of the monitor (Figure 1) to define the game scene was divided by the game situations in the monitor screen and calibrate the gaze position. With solo rank game of LoL. According to previous research, calibration, the coordinates of the participant's gaze esports players have a wide gaze distribution when position were represented as a number between 0 and performing multiple tasks simultaneously [4,10]. 1. The participants used their own mouse and Therefore, we set the two hypotheses. H-1) Game keyboards for the experiment. situations that require multiple tasks have wider gaze distribution than not require multiple tasks. H-2) High-skilled players show wider gaze distribution than middle-skilled players in game situations that require them to perform multiple tasks simultaneously. 2. Methods 2.1. Participants Eight high-skilled and eight middle-skilled LoL players were recruited for the experiment. The official rank of the high-skilled players was over than platinum rank (top 10%). Middle-skilled players were involved in Figure 1: Surface markers and experimental bronze, silver, and gold tiers (lower than the top 10%). environment. Each red box indicates the surface All participants self-reported that they had normal marker used for detecting the monitor screen and vision with no gaming disorder. Table 1 shows specific calibrating the gaze movement. information about the participants. The experiment was approved by the Human Research Ethics Committee of The University of Tokyo (approval number: 872). 2 2.3. Experimental procedures Participants could move their characters by using the mouse right-click. In the Fighting scene, the participant Before the experiment, participants wore the eye freely fought with enemy team players by combining tracker and calibrated the gaze position using the the mouse left click and keyboard q, w, e, and r keys Screen Marker Calibration method. The Screen Marker (Figure 4B). Four different types of fighting scenes Calibration method calibrates the gaze position were included in the Fighting scene (Figure 4B-1, 2, 3, through five dots that appear on the screen (one center and 4). In the Object scene (Figure 4C), participants and four corners of the monitor) (Figure 2). fought with six types of objects that were operated by a computer AI system (Tower, Nexus, Dragon, Inhibitor, Rift Herald, and Baron). Participants can obtain items and buffs that are important in the game by defeating the six types of objects. Participants could destroy the objects by using the mouse left click and Figure 2: Overall flow of the “Screen-Marker keyboard q, w, e, r keys. The Watching scene included Calibration” method. the action of watching another player play to check the overall flow of the match by using the keyboard's left, When the calibration was finished, participants right, up, and down keys or mouse right-click (Figure logged in to their own LoL account for a solo-rank 4D). Finally, the ALL scene was defined as a game scene match. Before starting the task, we set the distance that was non-classified. The Assignment was finished between the monitor and the head position of the when the participants won or lost. participants as 100 cm. After setting the head position, we After the Assignment was finished, two parameters requested the participants to keep their current head were calculated to evaluate the performance level. The position as same as possible during the task. During the first is the Kill/Death/Assistant ratio (KDA) used to experiment, participants freely played the solo-rank match evaluate the performance level of each participant. (called the Assignment). Solo-rank match was designed KDA was calculated as following equation (1). by the publisher of LoL (Riot Games, Inc., California, USA). During the solo-rank match, participants teamed up 𝑒𝑙𝑖𝑚𝑖𝑛𝑎𝑡𝑒 𝑡ℎ𝑒 𝑒𝑛𝑒𝑚𝑦 + 𝑎𝑠𝑠𝑖𝑠𝑡𝑎𝑛𝑡 (1) with four random players to play the match (one team with 𝑒𝑙𝑖𝑚𝑖𝑛𝑎𝑡𝑒𝑑 𝑏𝑦 𝑒𝑛𝑒𝑚𝑦 five members). When teaming up with four random The second is “Total Damage to Champions” used players, it will be matched with players who have similar to evaluate the Assignment performance. "Total rankings to the participants by the AI matching system. Damage to Champions" represents the amount of Participants fought against the other opponent team direct damage to the opponent team characters. players (not a bot) who had similar ranks to them. When the match was finished, gaze movement was analyzed by open-source software for Pupil-Players 2.5. Gaze movement acquisition (version 3.5.7). The overall flow of the experiment can be checked in Figure 3. At the end of the Assignment, gaze distribution, Region of Interest (ROI), and fixation duration were calculated for each of the five scenes (Moving, Fighting, Object, Watching, and ALL scene). According to the manufacturer, it is recommended to only use data with a confidence level of 80% [13]. The confidence level was used as the accuracy of the gaze movement data. The confidence level was calculated by the accuracy of the pupil detected through the eye camera. A total of 7.18% of the gaze movement data (have lower Figure 3: Overall flow of the experimental procedure. accuracy than 80%) were excluded from the evaluation. In gaze distribution, the standard deviation 2.4. The Assignment of horizontal and vertical gaze was calculated respectively. The coordination of gaze position was The game scene in the Assignment was divided into normalized by the monitor size and represented as a four categories (Moving, Fighting, Object, and number between 0 to 1. We set the five following ROIs: Watching) and non-divided scene (ALL; total 5 game Chatting, Skill, KDA, Mini-map, and Game scene areas scenes). In esports, the game scene was classified (Figure 5). Participants could view the chat in the based on the game situation that commonly appeared Chatting area (Figure 5A). In the Skill area, participants during the gameplay [14]. Thus, we categorized the could see the remaining time of skill, purchased items, game scenes that commonly appeared during the LoL virtual commodity, and level of skills (Figure 5B). The single-rank matches as previous research. During the KDA area showed how many enemies a participant had experiment, we record the feature of the Assignment taken down and helped other teammates (Figure 5C). as a video .mp4 file without including the gaze In the Mini-map area, participants could see the entire movement data. After the experiment, a video file of flow of the game (Figure 5D). The Game scene area the Assignment was provided for an anonymous LoL represents the main screen where the Assignment was player who judged and divided the game scenes. The performed (Figure 5E). The percentage of gaze game scene was classified by the top 1% of the ranked position located in each ROI was calculated. Fixation anonymous LoL players with more than 10 years of was defined when the gaze position was fixed more experience. In the Moving scene, the participant only than 100ms and the maximum pupil dispersion was moves their character in the game (Figure 4A). less than 1.5 degrees. 3 Figure 4: Feature of the Assignment game scene. Each picture indicates the scene of the Assignment. A) Moving scene. The Moving scene includes the simple movement of the participant's character. B) Fighting scene. B-1,2,3,4 indicates the sub-category of the Fighting scene. The Fighting scene includes the battle of the participant character. C) Object scene. The Object scene includes when a participant hits the objects. D) Watching scene. The Watching scene includes when the participants watch the other players play. According to Levene’s test and Shapiro-Wilk test, all datasets did not follow the normality and homogeneity. Therefore, the Wilcoxon rank-sum test was performed to determine the difference in performance levels (KDA and “Total Damage to Champions) and experienced years between the groups (high- and middle-skilled players). Gaze movements (horizontal gaze distribution, vertical gaze distribution, ROI, and fixation duration) were analyzed by two-way analysis of variance (ANOVA) with aligned rank transform (ART), non-parametric statistical Figure 5: Each position of ROI. Each red box indicates methods [16], for two skill levels (high-skilled, middle- the ROI of the game. A) Chatting area, B) Skill area, C) skilled) and five scenes (Moving, Fighting, Object, KDA area, D) Mini-map area, E) Game scene area. Watching, ALL scenes). When a main effect was observed in scenes, a contrast test was performed for 2.6. Statistical analysis multiple comparisons. When a significant interaction between skill level and scenes was observed, the contrast test was performed as the post-hoc test. P All statistical analyses were performed by the RStudio values were adjusted by using the Holm-Bonferroni version 4.3.1 (R Studio, Boston, MA, USA). After the correction method. Partial η2 indicated effect size for experiment, a power analysis was conducted to the ANOVA. All levels of statistical significance were estimate the number of participants was appropriate set at p < .05. (G*Power version 3.1.9). Three parameters were used to estimate the power of the sample size. Horizontal gaze distribution in the Moving scene and the 3. Results Watching scene, ROI percentage between high- and middle-skill in the Mini-map area, and fixation duration between high- and middle-skill were used for 3.1. Scene classification result power analysis. The effect size was calculated according to Cohen’s method [15]. The specific results There was no significant difference between high- and of the power analysis can be checked in Table 2. middle-skilled players in Assignment playtime (high- skilled: 1790 sec. ± 328 sec., middle-skilled: 1479.4 Table 2 sec. ± 626 sec.; Wilcoxon rank-sum test, W = 17, p = Results about power analysis. A) Horizontal gaze .42). A total of 1214 scenes were classified from all distribution between the Moving scene and the Fighting participant’s game scenes (Moving: 358, Fighting: 419, scene. B) ROI percentage between high- and middle-skill Object: 265, Watching: 172). There was no significant in the Mini-map area. C) Fixation duration between high- main effect observed in skill level (F = 0.08, p = .77, and middle-skill. Partial η2 = .008). The main effect was detected in the Parameter P value F value Cohen’s Actual scene (F = 3.54, p = .02, Partial η2 = .31). According to d power the contrast test, the Watching scene had a smaller A) .01 5.68 1.61 0.83 number than the Fighting scene (p = .03). However, B) > .001 23.23 1.08 0.50 there was no significant difference in other scene compare results (Moving-Fighting: p = .97; Moving- C) .007 7.63 0.41 0.42 4 Object: p = .22; Moving-Watching: p = .09; Object- Fighting: p = .10; Watching-Object: p = .96). There was no interaction effect detected between skill level and scene (F = 0.49, p = .73, Partial η2 = .01). 3.2. Performance level The high-skilled players had a longer LoL experience than the middle-skilled players (Wilcoxon rank-sum test, W = 53.5, p = .02). Figure 6 indicates the performance level of each group. The high-skilled players had a significantly higher KDA than the middle-skilled players (Wilcoxon rank-sum test, W = 52, p = .04). Moreover, the high-skilled players had better “Total Damage to Champions” scores on average (high-skilled players: 21567 ± 5328.6, middle-skilled players: 14338.8 ± 14320.2, respectively). However, there was no significant difference in “Total Damage to Champions” scores between the high- and middle- skilled players (W = 48, p = .10). Figure 7: Horizontal and vertical gaze distribution. A, C: Difference between high- and middle-skilled players Figure 6: Performance level of high- and middle- in horizontal and vertical gaze distribution. The box skilled players. A: KDA of high- and middle-skilled plot indicates the average of high- and middle-skilled players. The bars indicate the average of KDA. The players. Whisker shows the standard deviation of the whiskers represent the standard deviation of the bar average. B, D: Difference of horizontal and vertical gaze plots. B: average “Total Damage to Champions” of high- distribution in each scene. The dots indicate the and middle-skilled players. The box plots indicate the average of each data. Whiskers show the standard average of the “Total Damage to Champions”. Whisker deviation of each data. A significant level was *p < .05, ** p < .01, and ***p < .001. shows the standard deviation of the bot plot. A significant level was *p < .05. 3.4. ROIs 3.3. Gaze distribution Figure 8 represents the percentage of gaze movement Figure 7 shows the statistical analysis results in gaze in each ROI. Two-way ANOVA with ART revealed no distribution. In horizontal gaze distributions, significant main effects (skill level: F = 0.20, p = .65, significant main effects in skill level and scene were Partial η2 = .006; scene: F = 1.94, p = .11, Partial η2 = observed (skill level: F = 5.68, p = .01, partial η2 = .07, .08) and interaction (F = 0.29, p = .88, Partial η2 = .03) scene: F = 8.56, p < .001, Partial η2 = .32). Contrast test in the Chatting area. In the Skill area, no significant found that Moving scene had wide gaze distribution main effect (skill level: F = 0.11, p = .73, Partial η2 = .03; than Fighting scene and Object scene (both p < . 001). scene: F = 0.18, p = .94, Partial η2 = .01) and interaction The Fighting scene had a narrower gaze distribution (F = 0.17, p = .95, Partial η2 = .004) was detected. In the than the Watching scene and ALL scene (p = .002 and KDA area, no significant main effect (skill level: F = p = .01, respectively). Finally, the Object scene had a 0.06, p = .79, Partial η2 = .001; scene: F = 0.30, p = .87, narrower gaze distribution than the Watching scene Partial η2 = .04) and interaction (F = 0.20, p = .93, and ALL scene (p = .005 and p = .02, respectively). Partial η2 = .04) was observed. In the Mini-map area, There was no significant interaction effect detected (F the main effect was detected between high- and = 0.27, p = .89, Partial η2 = .01). In vertical gaze middle-skilled players (F = 23.23, p < .001, Partial η2 = distribution, there were no significant main effects and .23), but not in scene (F = 0.40, p = .80, Partial η2 = .01). interaction (skill level, F = 0.97, p = .32, partial η2 = .01; No interaction was detected in Mini-map area (F = scene, F = 2.26, p = .07, partial η2 = .11; interaction, F = 1.11, p = .35, Partial η2 = .02). There was no main effect 0.11, p = .97, Partial η2 = .007). (skill level: F = 0.44, p = .50, Partial η2 = .01; scene: F = 0.33, p = .85, Partial η2 = .003) and interaction (F = 0.29, p = .87, Partial η2 = .004) in the Game scene area. 5 4. Discussion In the current study, we investigated the differences between high- and middle-skilled players' gaze movements depending on the situation. The Assignment of the current study (solo-rank game) had participants fight against opponent players in a motivated situation (affecting the participant’s official ranking). Moreover, participants fought against the human opponent players. The motivated situation and human opponent players allow the experiment to reveal the source of the high performance of LoL players in actual game situations. In terms of performance level (KDA and Total Damage to Champions), the high-skilled players had significantly higher performance levels (KDA) than the middle- skilled players (Figure 6A). Moreover, the high-skilled players had more experienced years than the middle- skilled players. This result indicates that each group (high- and middle-skilled players) was clearly divided, and high-skilled players had higher performance Figure 8: Percentage of gaze position in each ROI. A, C, levels than middle-skilled players. However, there was E, G, I: Box plots indicate the average percentage of no significant difference observed in Total Damage to ROI. Whiskers indicate the standard deviation of each Champions between the groups. Total Damage to boxplot. B, D, F, H, J: The dots indicate the average of Champions is affected by not only the individual each data. K: location of each ROI. Whiskers indicate performance but also the items and positions that the standard deviation of each boxplot. A significant participants used. For example, if participants select level was ***p < .001. the item and position to help the teammate rather than directly fight with enemy players, Total Damage to 3.5. Fixation duration Champions is naturally decreased. According to the scene classification result (Result Figure 9 represents the duration of gaze fixation in 3.1.), there was no significant difference in the number each scene. Two-way ANOVA with ART found that the of each scene between high- and middle-skilled main effect between high- and middle-skilled players players. In addition, the number of the Fighting scene (F = 7.63, p = .007, Partial η2 = .04). There was no main was greater than the Watching scene. Thus, analyzing effect of the scene (F = 0.44, p = .77, Partial η2 = .002) the characteristics of the gaze movement as a whole and interaction (F = 0.06, p = .99, Partial η2 = .002). game without categorizing the situation is likely to bias the overall results due to factors related to the specific situations. According to the results about horizontal and vertical gaze distribution, the high-skilled players had a horizontally wider gaze distribution than the middle- skilled players (Figure 7A). It is well known that dividing the gaze movement into horizontal and vertical directions was common practice in esports studies [4,8,10]. Moreover, previous research points out that gaze distribution and performance level in esports have significant correlations [17]. Thus, analyzing the gaze movement of esports players by dividing the gaze distribution helps to understand the superior gaze control ability of LoL players. Since the monitor is a long object in the horizontal direction, no significant difference between high- and middle- skilled players was caused by the physically short Figure 9: Fixation duration of each scene. A: average length in the vertical direction. Moreover, the high- fixation duration between high- and middle-skilled skilled players had a shorter fixation duration than the players. The bar plot shows the average fixation middle-skilled players (Figure 9A). These results are duration in each scene. Whisker indicates the standard consistent with previous studies that have shown that deviation of each bar plot. B: average fixation duration skilled LoL and real-time strategy (RTS) game players between each Scene. The dots indicate the average of had wider gaze distributions and short fixation times each data. Whisker indicates the standard deviation of [4,10]. each bar plot. A significant level was **p < .01. Horizontally wide gaze distributions might be caused by the superior visual processing skills of high- skilled LoL players. Generally, the wide gaze distribution is beneficial for collecting information 6 from a wide area in cognitive tasks and esports be important factors in achieving the high [4,10,17,18]. Since the user interface and information performance of the high-skilled players in LoL. in LoL are widely spread on the monitor, it is essential However, there was no significant difference in the to check the entire screen for the information that LoL ROI percentage of Skill and KDA area between high- players need. Obtaining information not only from a and middle-skilled LoL players (Figure 8C and E). wide area but also quickly is important for achieving There is a possibility that both high- and middle- high visual processing skills. For example, short skilled players had superior visuo-spatial ability. The fixation time with high task accuracy in cognitive tasks visuospatial ability is the capacity to memorize and represents high visual processing skills [19]. In understand visual-spatial objects correctly [21]. Both fixation duration, high-skilled LoL players had a information in the Skill area and KDA area were shorter fixation duration than middle-skilled LoL related to the visual and spatial elements. In the Skill players regardless of the game scene. This indicates area, participants could check the left time of the skills. that getting information quickly, regardless of the In the KDA area, participants could see the information game situation, might be a superior characteristic of about time and KDA. Previous research points out that high-skilled LoL players. To sum up, short fixation time long-term esports training can improve visuospatial and horizontally wide gaze distribution indicate that ability [22]. In the current research, all participants high-skilled LoL players had superior visual had enough LoL experience (experienced years; high- processing skills than middle-skilled players. skilled: 9.2 years, middle-skilled: 4.3 years). Thus, it Surprisingly, horizontal gaze distribution changed might be able to estimate the information without significantly depending on the game situation seeing the Skill and KDA area with high visuospatial regardless of the skill level (Figure 7B; H-2). However, ability. Thus, each participant had enough experience there was no significant difference between high- and to guess what was being displayed without looking middle-skilled LoL player’s gaze movements in specific directly at the information on the Skill and KDA area. scenes (H-1). The results related to H-1 show that LoL In the Chatting area, there was no significant players should be able to control their gaze difference between high- and middle-skilled players movements for specific situations, regardless of their detected in ROI percentage. The reason is as follows: skill level. The Moving scene had a wider horizontal Both high- and middle-skilled players did not prefer to gaze distribution than the Fighting and Object scene. use the Chatting area because typing the chatting takes Participants must move their characters after a long time to communicate with other players. It is understanding the overall flow to win the game. To important to reduce wasting time to react fast during process the visual stimulation, it is necessary to the game. Therefore, there is a possibility that visually check the target first [20]. Thus, it is beneficial participants spent less time communicating with other for winning the game to distribute the gaze movement players by using the feedback icons rather than typing and gathers a broader range of information than in the a chatting. Fighting and Object scene through the wide gaze The Game scene area is the main area where the movement. Fighting scenes and Object scenes have a game is played. Therefore, the Game scene area had a narrower distribution of gaze than Watching and ALL high importance in both high- and middle-skilled scenes since more information is concentrated in the players. High importance might be effect to no center of the monitor. Furthermore, the gaze control significant difference in the Game scene area was ability observed throughout the game (ALL scene) is observed between high- and middle-skilled players. not the same as that in the Fighting scene. Previous research suggests training to widen the gaze distribution simply since skilled esports players have 5. Limitation a wider gaze distribution during games [4]. However, according to the result of the current study, it is In the current study, we only experimented with necessary to conduct situation-based gaze control eight high-skilled and eight middle-skilled LoL players training in LoL players regardless of their skill level. (small sample size). Thus, some parameters have low In the ROI percentage of gaze position, high-skilled power which is related to sample size (see Table 2). It players had a higher ROI percentage in the Mini-map is important to be cautious about applying the area than middle-skilled players for two reasons obtained results to all LoL players. In future research, (Figure 8G). First, the game interface is designed to conducting the experiment with a large sample size is share the information in the Mini-map area. For necessary. To induce the actual solo-rank gameplay example, participants were able to give feedback on situations, we did not strictly control the trial of each dangers (e.g., enemy is coming) with icons in the Mini- scene. Thus, there is a possibility that some scenes map area. Second, the overall flow of the Assignment might have more or fewer gaze points and affect the was represented in the Mini-map area. Understanding gaze movement-related data. Moreover, we did not the overall flow of the game and cooperating with strictly control the head movement. Therefore, we teammates are key factors to winning. According to cannot exclude the possibility that head movement previous research about RTS game players, high- affected the gaze movement data. skilled RTS game players frequently check the overall flow of the games than low-skilled players [4]. RTS game and LoL had similar user interfaces (e.g., the 6. Conclusion mini-map was represented in the right corner of the monitor). Thus, focusing the gaze position on the The current study analyzed the gaze control ability information about feedback and the overall flow might of esports players in various and motivated situations. High-skilled LoL players were advantageous to collect 7 the information from a wide area through a wider gaze Skill Levels”, 2023 International Workshop on Smart distribution and a shorter gaze fixation time than Info-Media Systems in Asia (SISA 2023), Aug.31- middle-skilled players regardless of the game situation. Sep.1. Since overall flow and communication information had [9] M. Mora-Cantallops and M.-Á. Sicilia, “MOBA an essential role in achieving high performance, high- games: A literature review,” Entertain. Comput., skilled players saw the area displaying overall flow vol. 26, pp. 128–138, 2018, doi: and communication information than middle-skilled https://doi.org/10.1016/j.entcom.2018.02.005. players. Surprisingly, the gaze control abilities showed [10] I. Jeong, K. Kudo, N. Kaneko, and K. Nakazawa, significant differences between full games and specific “Esports experts have a wide gaze distribution situations, regardless of skill level. Specifically, gaze and short gaze fixation duration: A focus on distribution was wider in game situations, which League of Legends players,” PLoS One, vol. 19, no. require focusing on multiple pieces of information 1, p. e0288770, Jan. 2024, [Online], doi: simultaneously, than in other situations. Therefore, https://doi.org/10.1371/journal.pone.0288770. esports research should investigate the gaze control [11] D. Railsback and N. Caporusso, “Investigating the ability of esports players in separate game situations Human Factors in eSports Performance BT - might shed light on the development of the training Advances in Human Factors in Wearable method in esports. Technologies and Game Design,” 2019, pp. 325– 334. [12] F. Brühlmann, P. Baumgartner, G. Wallner, and S. Acknowledgments Kriglstein, “Motivational Profiling of League of Legends Players,” vol. 11, no. July, pp. 1–18, 2020, We appreciate Minjun Kim for providing information doi: 10.3389/fpsyg.2020.01307. on LoL-related knowledge. The current research was [13] M. Kassner , W. Patera , A. Bulling , “Pupil: An supported by the Tateisi Science and Technology Open Source Platform for Pervasive Eye Tracking Foundation (grant number: 2237004), Meiji Yasuda and Mobile Gaze-Based Interaction”, Proceedings Life Foundation of Health and Welfare. of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct Publication, 2014, 1151-1160, doi: References https://doi.org/10.1145/2638728.2641695. [14] Y. Maeda, “Scene Classification in FPS Game [1] V. H. S. E. Robertson, “eSports and Digital Videos Based on Acoustic Information.” 2023 Ecosystems: An Antitrust Perspective,” J. Eur. IEEE Conference of Games, 2023, Compet. Law Pract., vol. 12, no. 8, pp. 591–592, ISBN: 9798350322774 Oct. 2021, doi: 10.1093/jeclap/lpab040. [15] J. Cohen, Statistical Power Analysis for the [2] J. G. Reitman, M. J. Anderson-Coto, M. Wu, J. S. Lee, Behavioral Sciences. 2013. and C. Steinkuehler, “Esports Research: A [16] J. O. Wobbrock, L. Findlater, D. Gergle, and J. J. Literature Review,” Games Cult., vol. 15, no. 1, pp. Higgins, “The Aligned Rank Transform for 32–50, 2020, doi: 10.1177/1555412019840892. Nonparametric Factorial Analyses Using Only [3] M. W. G. Dye, C. S. Green, and D. Bavelier, Anova Procedures,” in Proceedings of the SIGCHI “Increasing Speed of Processing With Action Conference on Human Factors in Computing Video Games,” Curr. Dir. Psychol. Sci., vol. 18, no. Systems, 2011, pp. 143–146, doi: 6, pp. 321–326, Dec. 2009, doi: 10.1111/j.1467- 10.1145/1978942.1978963. 8721.2009.01660.x. [17] I. Jeong, T. Nobuto, N. Kaneko, T. Kato, and K. [4] J. Inhyeok, K. Nakagawa, R. Osu, and K. Kanosue, Nakazawa, “Investigating the gaze control ability “Difference in gaze control ability between low of VALORANT players using a Python based tool,” and high skill players of a real-time strategy arXiv Prepr., doi: game in esports,” PLOS ONE. pp. 1–17, 2022, doi: https://doi.org/10.48550/arXiv2310.15542, 10.1371/journal.pone.0265526. 2023. [5] W. R. Boot, A. F. Kramer, D. J. Simons, M. Fabiani, [18] M. Sanz-matesanz, G. M. Gea-garcía, and L. M. and G. Gratton, “The effects of video game playing Martínez-aranda, “Computers in Human on attention, memory, and executive control,” Behavior Physical and psychological factors Acta Psychol. (Amst)., vol. 129, no. 3, pp. 387–398, related to player ’ s health and performance in 2008, doi: 10.1016/j.actpsy.2008.09.005. esports : A scoping review,” Comput. Human [6] A. Klostermann, C. Vater, R. Kredel, and E.-J. Behav., vol. 143, no. August 2022, p. 107698, Hossner, “Perception and Action in Sports. On the 2023, doi: 10.1016/j.chb.2023.107698. Functionality of Foveal and Peripheral Vision,” [19] S. Pannasch, J. Schulz, and B. M. Velichkovsky, “On Front. Sport. Act. Living, vol. 1, 2020, doi: the control of visual fixation durations in free 10.3389/fspor.2019.00066. viewing of complex images,” Attention, [7] X. Li, L. Huang, B. Li, H. Wang, and C. Han, “Time Perception, Psychophys., vol. 73, no. 4, pp. 1120– for a true display of skill: Top players in League 1132, 2011, doi: 10.3758/s13414-011-0090-1. of Legends have better executive control,” Acta [20] B. D. Cameron and J. López-Moliner, “Target Psychol. (Amst)., vol. 204, p. 103007, 2020, doi: modality affects visually guided online control of https://doi.org/10.1016/j.actpsy.2020.103007. reaching,” Vision Res., vol. 110, pp. 233–243, [8] F. Ryousuke, and H. Goichi, “Gaze and 2015, doi: Electroencephalography (EEG) Parameters in https://doi.org/10.1016/j.visres.2014.06.010. Esports: Examinations Considering Genres and 8 [21] F. Irani, “Visual-Spatial Ability BT - Encyclopedia of Clinical Neuropsychology,” J. S. Kreutzer, J. DeLuca, and B. Caplan, Eds. New York, NY: Springer New York, 2011, pp. 2652–2654. [22] L. Milani, S. Grumi, and P. Di Blasio, “Positive Effects of Videogame Use on Visuospatial Competencies: The Impact of Visualization Style in Preadolescents and Adolescents,” Front. Psychol., vol. 10, 2019, doi: 10.3389/fpsyg.2019.01226. 9