Quantifying Wrist-Aiming Habits with A Dual-Sensor Mouse: Implications for Player Performance and Workload

Computer mice are widely used today as the primary input device in competitive video games. If a player exhibits more wrist rotation than other players when moving the mouse laterally, the player is said to have stronger wrist-aiming habits. Despite strong public interest, there has been no affordable technique to quantify the extent of a player’s wrist-aiming habits and no scientific investigation into how the habits affect player performance and workload. We present a reliable and affordable technique to quantify the extent of a player’s wrist-aiming habits using a mouse equipped with two optical sensors (i.e., a dual-sensor mouse). In two user studies, we demonstrate the reliability of the technique and examine the relationship between wrist-aiming habits and player performance or workload. In summary, player expertise and mouse sensitivity significantly impacted wrist-aiming habits; the extent of wrist-aiming showed a positive correlation with upper limb workload.


INTRODUCTION
Since its first release to the public in 1968, computer mice have been used as a key peripheral for human-computer interaction (HCI) in the desktop environment [2,6,40].Although new interaction paradigms such as mobile [3] and virtual reality (VR) [17] have emerged, they are still widely used and continue to be developed, perhaps due to the rapid growth of multiplayer online PC games in desktop environments.Millions of people are accessing online games of various genres, such as first-person shooters (FPS) [20,39], real-time strategy (RTS) [11], and multiplayer online battle arenas (MOBA) [27], within which fierce competitive battles take place every day with a computer mouse as a main weapon (a.k.a., esports).Players hoped to have a more accurate, precise, and faster computer mouse than others to win the game, and such a demand motivated the continuous development of the high-performance computer mouse.The sensing precision of a mouse equipped with an optical sensor has been improved hundreds of times compared to earlier versions [46].With a current state-of-the-art commercial computer mouse [1], we can robustly measure hand movement on a desk surface with a resolution of up to approximately 0.00071 mm (=36,000 counts per inch, CPI).
In addition to improving mouse hardware specifications led by manufacturers, people are also actively researching the optimal use of a computer mouse.For example, previous studies discussed whether variables such as mouse sensitivity [5], weight [13,36,53,55], shape [47], connection type [53], and friction of the mouse pad [50] affect game performance and, if so, how to properly set them.However, perhaps the most actively discussed issue concerning optimal mouse usage in video games is a player's wrist-aiming habits.Here, aiming (more generally, aimed movement) refers to the general act of moving a specific end effector on the screen (i.e., pointer or crosshair) to another location by controlling it with a computer mouse [4,16,44].When moving the mouse horizontally for aiming, the rotation of the human upper limb joints inevitably follows (i.e., wrist, elbow, and shoulder) [32,33], and it can be said that any player who rotates the wrist joint more per unit mouse displacement has a stronger wrist-aiming habit (or a weaker armaiming habit).With a web search, we found 50,500 posts related to wrist-aiming habit1 , and one video among them got 815,000 views [29], showing players' high interest in this issue.
By further analyzing the top 100 posts and 25 videos on the web related to wrist-aiming habits, we looked at the specific speculations that existed about the impact of wrist-aiming habits on players 2 .Interestingly, 61 % of the posts and videos recommended arm-aiming rather than wrist-aiming.As reasons for the recommendation, 50 % of them cited improved aiming performance, 18 % cited prevention of wrist injuries, and the remaining 32 % did not provide any specific reasons.More specifically, 32% of them claimed that arm-aiming is associated with low mouse sensitivity, which allows for more stable aiming.On the other hand, only 11 % of all posts and videos recommended wrist-aiming, and 71 % of them also cited improved aiming performance as the reason.Only 8 % of the posts claimed that which aiming method to use should vary depending on the situation.In summary, players today seem to believe that aiming habits have a significant impact on player performance and the likelihood of injury.However, despite the High W-index Low W-index Lateral mouse speed lo w W -in d e x h ig h W -in d e x

Wrist angular velcotiy
Figure 1: A player's wrist joint angular velocity   can be reliably estimated from a dual-sensor mouse without using special equipment such as motion capture cameras.In this study, we quantify the strength of players' wrist aiming habits (i.e., W-index) from the relationship between mean wrist angular velocity  [   ] and mean lateral speed of the mouse body  [   ] measured from a dual-sensor mouse.
high demand from players, the guidelines on the web are either conflicting or not sufficiently supported by scientific evidence.
We believe that the main reason why there are only guesses and no scientific conclusions about wrist-aiming habits is the lack of robust and affordable technology to quantify the extent of wristaiming habits.Lacking objective quantification, we rely on subjective interpretation, hindering consensus on meaningful scientific conclusions.Motion capture systems can be used to analyze wristaiming habits by measuring the rotation of each arm joint as a player moves the mouse, but this method is not affordable for most players and teams.In this study, we propose a novel technique to quantify a player's wrist-aiming habits using a relatively affordable device, a dual-sensor mouse [30,33], and based on the technique, we broadly investigate the relationship between wrist-aiming habits, performance, and upper limb workload.The dual-sensor mouse is a special mouse with two optical sensors and can measure the angular velocity of the mouse body   , unlike ordinary mice.From   measured in a dual-sensor mouse, the angular velocity of a player's wrist joint   can be estimated reliably, even without expensive equipment such as motion capture cameras (to be verified in Sections 4 and 5).The recipe for making a dual-sensor mouse is published as an open source, and it only costs about 100 USD.
More specifically, two pieces of information measured from a dual-sensor mouse are used to quantify a player's wrist-aiming habits: (1) mean angular velocity of wrist joint  [   ] and (2) mean lateral velocity of mouse  [   ].A scatter plot of all ( [   ],  [   ]) points measured during gameplay is called a W-diagram (see Figure 2b), and the slope of the linear regression line for those points is the W-index, which represents the strength of the player's wrist aiming habits.The higher the player's W-index, we can interpret that the player rotated the wrist joint faster on average relative to the implemented mouse speed; that is, the player showed stronger wrist-aiming habits on average (Figure 1).
Two user studies were conducted to verify the reliability of the proposed technique and address questions about the impact of wrist-aiming habits on player performance and workload.In Study 1, we tested whether the mean angular velocities of players' wrist joints could be reliably estimated using only a dual-sensor mouse ( =20).We compared mouse angular velocities measured from a dual-sensor mouse to wrist angular velocities measured from a motion capture system.In Study 2, we recruited amateur and professional first-person shooter (FPS) players ( =17) and looked at how their W-index, upper limb workload, and task performance varied as they performed a wide variation of aim-and-shoot tasks under different mouse sensitivities.Upper limb workload was inferred through forces measured from a sensor placed under the mouse pad.Findings from user studies can be summarized as follows: • Average wrist angular velocity can be estimated with a high coefficient of determination ( 2 >0.94) across three different game genres, multiple participants, and a wide mouse sensitivity range using only a dual-sensor mouse.• The proposed technique allowed us to reliably quantify a player's wrist-aiming habits (average  2 ≈0.99).• Professional FPS players showed an average W-index 20.2% higher than amateurs (i.e., stronger wrist-aiming).• In general, W-index increased as mouse sensitivity increased.
• The average force applied to the mouse pad, which represents the workload imposed on the upper limb, showed a positive correlation with the W-index.• The effect of aiming habits on the aim-and-shoot performance in FPS appears to be weak; Performance and W-index showed only weak positive correlations, limited to some players.
The key contribution of this study is the presentation of a novel, low-cost technique that can reliably quantify the extent of a player's wrist-aiming habits.The technique allowed us to conduct quantitative research on the relationship between wrist-aiming habits, player performance, and upper limb workload, providing scientific evidence or refutations for claims floating around the web.We believe our study can provide useful insights regarding performance and health for amateur and professional players who play games for hours every day.To maximize impact, all technologies will be released as open source3 .

(a) (b)
Figure 2: (a) At a particular time , when the player is using a mouse, the angular velocity of the wrist joint, lateral velocity, and longitudinal velocity of the mouse are represented as   (),   (), and   (), respectively.(b) By observing a player's mouse control process, we can draw a scatterplot of the mean lateral velocity of the mouse  [  ] and the mean angular velocity of the wrist  [   ].This plot is called the player's W-diagram in this study.The W-index, which quantifies the strength of the player's wrist-aiming habits, is defined as the slope of the linear regression line in the W-diagram.

RELATED WORK 2.1 Ergonomic Issues in Mouse Use
To our knowledge, there have been no previous studies that specifically explored how mouse use posture affects the workload on the upper extremities.Meanwhile, several previous studies have shown that regardless of posture, prolonged mouse use generally places a higher workload on the upper extremities.IJmker et al. reported that people who use a mouse for at least 4 hours a day at work are at slightly increased risk of neck/shoulder pain symptoms [23].Madeleine et al. revealed that excessive use of a mouse is a major risk factor for forearm pain [42], and Keir et al. found that long-term mouse use can increase the risk of median mononeuropathy [26].Kang et al. explained that when using a mouse, if pressure is concentrated on the wrist and pisiform, it can cause various pathologic conditions on the ulnar side [25].They also mentioned that if such abnormal contact pressure is repeatedly applied, musculoskeletal pains of the forearm, wrist, or hand may develop depending on the strength and duration of mouse use [25].Ergonomic problems like these are more pronounced in hardcore gamers who use computers and mouse more frequently.According to interviews and surveys conducted regarding injuries to esports players in 2022, many esports players have experienced a decline in their skills due to injuries, and some have had to retire early due to severe injuries [35].Representative players who retired early due to injuries include League of Legends esports player Hai "Hai" Lam, Liu "Mlxg" Shi-Yu, and Kurtis "Toyz" Lau Wai-Kin.DiFrancisco-Donoghue et al. reported that over 30% of 65 college esports players from the United States and Canada experience hand and wrist pain [15].Another recent study found that 41 of 153 collegiate esports athletes in the United States had at least one type of upper extremity injury, and three of them even required surgery [12].Emara et al. recommended a wrist support mouse pad and full wrist exercises to prevent carpal and ulnar tunnel syndrome in esports players [19].

Better Use and Design of A Computer Mouse
Issues such as what posture the mouse should be used in and what is the appropriate design of mouse hardware and software have been mainly studied to improve players' input performance rather than reduce the workload on the upper extremities.Park et al. [49] used motion capture to track the upper limb movements of professional and amateur players in a first-person shooters (FPS) game and found that professionals rotated their wrists less during play than amateurs.This finding is in line with the negative opinion about the wrist-aiming habit on the web.However, in the study, the professional players set their mouse sensitivity much lower than the amateurs and consequently had to move their hands farther during the game, so they may have relied more on elbow rotation than wrist rotation.We do not yet know if professional players still show weaker wrist-aiming habits than amateurs at the same mouse sensitivity or how a player's wrist-aiming habits and input performance are affected by varying sensitivity over a wider range.
Regarding the air mouse, which is a mouse that controls the pointer by moving the arm in the air without placing the hand on the desk, several studies [10,38] have examined how input performance and fatigue are affected depending on which joint users mainly use during control.However, aimed movement in the air differs qualitatively from the desktop mouse control process; Air mouse control places a much higher workload on the shoulder and is not generally utilized in tasks that require high performance.
Among several variables related to mouse design that affect user input performance, the control-display (CD) gain or sensitivity function of a computer mouse has been studied most intensively.Researchers found that a gain that increases with pointer speed (i.e., pointer acceleration) rather than a constant one enables shorter pointing time [8], which is applied to most operating systems today [7].A guideline for determining an appropriate gain value for a computer mouse [8] or a technique for gradually optimizing and personalizing the shape of a gain function has also been proposed [34].On the other hand, competitive video game players are known to turn off pointer acceleration and use a constant gain function [5].A recent study found that mouse sensitivity that was too high or too low had a detrimental effect on player performance [5].
A dual-sensor mouse equipped with two optical sensors has triggered several studies on optimal mouse settings because it allows the angular velocity of the mouse body to be measured.In 2015, Lee et al., through a study using a dual-sensor mouse, showed that mouse input performance can be significantly improved when mouse coordinate disturbance is eliminated [33].In 2020, Kim et al. found through another study using a dual-sensor mouse that the location of the optical sensor at the bottom of a typical mouse has a significant effect on input performance [30].They open-sourced the recipe for their dual-sensor mouse implementation 4 , which was utilized in our study.
Today's gaming mice come in huge variations in weight, shape, connection type (wireless vs. wired), and more.However, whether such differences have a significant effect on input performance has only recently begun to be studied.Some of recent studies [37,53] reported that mouse weight or connection type has little effect on input performance, but most professional players prefer lightweight wireless mice.On the other hand, Conroy et al. [13] reported that participants achieved 4 % higher speed and 9 % higher accuracy in a target acquisition task when using a 50-, 60-, or 90-gram mouse compared to a 100-gram mouse.Li et al. [36] also reported that using an 80-gram mouse during a pointing task resulted in significantly lower path deviations than using an 87-gram mouse.Yan et al. [55] reported that an overly light mouse (40 grams) can increase the pointing error rate and that a mouse weighing about 60 grams is appropriate for both performance and workload.Odell et al. [47] found that an ordinary flat mouse is advantageous for obtaining high pointing performance compared to a vertical mouse, but requires a less ergonomic posture (i.e., more forearm pronation).This explains why vertical mice are rarely used among competitive gaming players, where achieving high performance is paramount.

Summary
In competitive video games, players generally prioritize achieving high performance above all other values.Therefore, previous studies on how to design a computer mouse and what posture to use a mouse have focused on their impact on player performance.However, as the gaming industry grows and the number of players who use the mouse excessively increases, the problem of upper extremity injuries caused by mouse use is also becoming more serious than before.To study the complex relationship between mouse usage posture, mouse design, upper extremity workload (or injuries), and input performance, what must be preceded is the development of a technology to standardly quantify a player's mouse usage posture (or habits), which has not been attempted in previous studies.

Kinematics of Mouse Control
Mouse control in this study refers to the process in which a player moves a computer mouse placed on a flat surface to a desired location without separating it from the surface.Mouse control is generally done in a seated position, and a player's upper limb holding the mouse can be regarded as a 7-degree-of-freedom (DOF) linkage with 7 revolute joints (see Figure 3a).For the sake of simplicity, in this section, we assume that the player is right-handed.
When looking down at the surface vertically, we assume that the center of the mouse always overlaps the center of the player's hand during mouse control.We also assume that the orientation of the mouse remains roughly parallel to the line connecting the player's wrist point and the center of the hand (see Figure 3b).In the vertical view, the player's upper limb can be further simplified as a 3 DOF linkage with three revolute joints: shoulder, elbow, and wrist [33].Let the angular displacements of each joint be   ,   , and   in the radian unit.If all three angles are zero, the player's arm extends horizontally to the right.The angular velocity of each joint is denoted as   ,   , and   , respectively.

Redundancy in Mouse Control
In the mouse control process, what players want to change and determine is the mouse's position on the surface.The mouse's position on the surface can be determined in two-dimensional coordinates,   and   .At this time, since the DOF of the upper limb linkage is higher than the DOF required to determine the mouse position, the mouse control process is said to be redundant.In other words, there can be infinitely many combinations of (  ,   ,   ) that allow the mouse to be positioned at a specific location (  ,   ) for a given shoulder position (  ,   ).Due to the redundancy in mouse control, wrist-aiming habits can vary from player to player.Some players have stronger wrist-aiming habits, which means that they resolve kinematic redundancy in a way that relies more on wrist rotation than other players, even when moving the mouse along the same trajectory (see Figure 3c).

Wrist-Aiming Habit Index
In this study, we propose a scalar index that correlates with how strong a player's wrist-aiming habit is.The index is named the wrist-aiming habit index (W-index) and is calculated from (1) the mouse velocity and (2) the angular velocity of the wrist.During mouse control, the velocity of the mouse can generally be directly measured, but the angular velocity of the wrist cannot be directly measured without special settings such as a motion capture system.In this section, we first describe how the W-index can be calculated when both are measured.The estimation of wrist angular velocity using a dual-sensor mouse is deferred to the following section.
Suppose that at a particular time , a player is moving a computer mouse on a surface with a velocity ì Here, it is assumed that the velocity is measured with respect to the mouse sensor coordinate system.That is,   refers to the lateral velocity of the mouse, and   refers to the longitudinal velocity component of the mouse.By simplifying the player's upper limb as a 3 DOF linkage as presented in Section 3.1, the angular velocity of the wrist joint is expressed as   () (see Figure 2a).
Next, assume that we observed the player's mouse control behavior for a sufficiently long period of time (e.g., = 1 to   ), collecting These pairs are then binned with respect to   .If there are  bins and the average of   and   for each bin is obtained, we can write: Here,  [] refers to the average or expected value operator.All calculated ( [  ],  [   ]) pairs can be drawn as a scatter plot, which we call a wrist-aiming habit diagram or a W-diagram (see Figure 2b).Since moving the mouse faster requires proportionally faster wrist rotation, the data points on the plot will lie roughly on a line passing through the origin with a positive slope.The slope of the line obtained through simple linear regression is the player's W-index that we finally propose.A higher W-index means that the player, on average, rotated the wrist joint more to move the mouse laterally.In other words, it can be interpreted that players with a high W-index have stronger wrist-aiming habits.For example, if a player's W-index is twice as high, it can be interpreted that the player makes wrist rotations on average twice as fast as other players in mouse control.
Note: The reason for drawing the W-diagram with bin averages rather than raw pairs of   and   is to average out the noise in   and   measurements and the variability of player behavior to enable a more robust estimation of the W-index.In other words, a higher W-index for a player simply means that that player's wrist rotates more on average.

Estimating Mean Wrist Angular Velocity 𝐸 [ 𝜃 𝑤 ] with A Dual-Sensor Mouse
The technique in the previous section assumes that two pieces of information are observed: (1) mean lateral velocity of the mouse The dual-sensor mouse is a special mouse with two optical sensors and has been utilized in HCI studies since 2015 [30,33,49].A recipe for self-manufacturing a dual-sensor mouse using an Arduino board and a 3D printer has been released as an open source [30].The sensors are installed along the central axis of the mouse (see Figure 4).Each of the sensors is called sensor 1 and sensor 2 in order of proximity from the mouse's buttons, and the -axis of all sensors is designed to overlap the central axis of the mouse.Count values measured from the sensors while the mouse moves are represented as ( 1 ,  1 ) and ( 2 ,  2 ), respectively.When the mouse is moved to the right or upward, a positive  1 or positive  1 is measured for sensor 1, and a negative  2 or negative  2 is measured for sensor 2, respectively (see Figure 4).
Unlike general computer mice, the dual-sensor mouse allows us to measure the amount of rotation of the mouse body.If the count readings from each sensor are ( 1 ,  1 ) and ( 2 ,  2 ) at time  when the mouse is moving on the surface, then the angular velocity of the mouse body   can be estimated as follows [33]: Here,   means the physical distance between the sensors (unit: inches),  refers to the sample time of the mouse sensor (unit: seconds), and  represents the sensitivity or CPI of the mouse (unit: counts/inch).The angular velocity of the mouse body is also positive in the clockwise direction.
Assuming that the mouse is properly fixed to a player's hand during mouse control, we can expect the amount of mouse rotation to show a strong correlation with the amount of rotation of the player's wrist on average.This proportional relationship is mathematically expressed as: Here,  is a constant of proportionality, which is generally expected to be smaller than 1, because not only the rotation of the wrist but also the rotation of the elbow and shoulder contribute.
Assuming that Equation 2 actually holds, we can obtain ( [  ], [   ]) pairs from the raw measurements of (  ,   ) and then apply Equation 2 to convert them into ( ).Therefore, the W-index can be estimated using only measurements from a dual-sensor mouse.In the following section, we validated our technique based on ground truth measurements from a motion capture system.

STUDY 1: TECHNIQUE VALIDATION
In this study, we validate that the mean wrist angular velocity  [   ] measured from motion capture can be reliably predicted from the mean angular velocity measured from a dual-sensor mouse  [   ] (see Equation 2).Validation is conducted across three different game genres and each participant's W-index is also examined.

Method
4.1.1Participants.A total of 20 participants were recruited from a local university (8 in their 20s and 12 in their 30s).All participants answered that they had played at least 100 matches of three different genres of games used in the experiment, and they were familiar enough to play the games immediately without any practice.Rank or tier within the games was not considered in the participant recruitment process.All participants were right-handed with no upper limb injuries at the time of participation.
4.1.2Task and Design.In this study, participants were asked to freely play each of the following three genres of games for more than 15 minutes using a dual-sensor mouse: • StarCraft (RTS) • Sudden Attack or Overwatch (FPS) • League of Legends (MOBA) The games are selected from famous and popular ones for each genre.In the case of the FPS genre, participants were allowed to choose a game that they had played at least 100 matches.If they have played more than 100 matches in both games, the game with the higher number of matches played was selected.The number of participants assigned to each FPS game resulted in: Sudden Attack ( =17), and Overwatch ( =3).StarCraft was played in 1 vs. 1 custom game mode against a computer bot.Sudden Attack, Overwatch, and League of Legends were played in Team Deathmatch mode (in Warehouse Map), 6 vs. 6 Quick Play mode, and 5 vs. 5 Howling Abyss mode with online players, respectively.Mouse sensitivity was allowed to be set freely by each participant for each game.The games were played in the order of MOBA, FPS, and RTS.
While participants were playing the game, we collected raw measurements from the mouse ( 1 ,  1 ,  2 ,  2 ) and the threedimensional positions (,,) of the participants' upper limb joints (shoulder, elbow, wrist) as measured from a motion capture system (13 cameras, 5 optical markers).The sampling rate of the mouse signal was 500 Hz, and the motion capture was 240 Hz.
4.1.3Procedure.Before participants arrived at the lab, we calibrated the motion capture system (see Figure 5a).As participants arrived, we gave them instructions for the experiment and they filled out a questionnaire and consent form.The participants set up the desktop environment where gameplay would take place comfortably (e.g., location of the mouse pad, height of the chair, distance to monitor).Then, five motion capture markers were attached to key locations of each participant's upper limb, including three joint locations (see Figure 5c).The participants were given time to adjust game-specific settings (e.g., mouse sensitivity, graphical settings) and practice to get used to the mouse.Then, they started to perform the main part of the experiment (free gameplay).Sufficient breaks were given between gameplay of different genres.Participants could stop participating at any time if they wanted.It took about an hour for each participant to complete all procedures, and a reward worth 30 USD was paid to each participant.).The dual-sensor mouse used in the experiment was constructed in 2022 according to the recipe on the web and the inside is composed of two optical sensors (PMW 3360, maximum sensitivity 12000 CPI) and an Arduino board (Spark Fun Pro micro).The weight of the mouse is 60 grams, which falls within the typical range of mouse weights (50 to 130 grams), and belongs to the category of lightweight mouse that people generally prefer [55].The distance between the sensors was 2.83 inches, and the sensor position of a dual-sensor mouse is simulated to be at the center of the two sensors [30].As a mouse pad, a general cloth surface type was used.A total of 13 Optitrack motion capture cameras were used (Prime  13, sampling rate 240 Hz).Dual-sensor mouse signals were collected through a Python logger, and motion capture data was collected using Optitrack's official API (https://optitrack.com/software/natnet-sdk/).Motion capture calibration was performed using OptiTrack Motive software (Motive 3.0.0Beta3).Both logging threads were triggered simultaneously with a single script.

Result
4.2.1 Pre-processing.There are parts of motion capture data and mouse data that are unrelated to wrist-aiming habit analysis, and we removed them through pre-processing.First, the delay between motion capture data and mouse data was estimated from crosscorrelation (30 ms), and the timestamp was unified.Second, rows with fewer or more motion capture markers were removed.Third, the markers in the motion capture were labeled based on the nearest neighbor algorithm and kinematic heuristic [21].Fourth, data from periods when the mouse was stationary or lifted from the surface (i.e., clutching [45]) were removed.Data 200 ms before and after clutching was also removed to account for instability.1 summarizes the mean speed at which participants moved the mouse in the lateral or longitudinal direction, the number of clutches per minute, the mean time spent per clutch, and the mean rotation angle of each joint for each game.From this, we can see that the participants' mouse control behavior significantly differed depending on the game genre.The participants' average arm lengths, as measured by the distance between motion capture markers, were: hand-to-wrist 4.4 cm (=0.8),wrist-to-elbow 24.2 (=0.9),elbow-to-shoulder 26.6 (=2.8).2), which is essential to quantify wrist-aiming habits using a dual-sensor mouse.Before the analysis, only data rows in which the lateral velocity component of the mouse was dominant were left (see Section 3.3).More specifically, only rows where the angle between the velocity vector (  ,   ) and the -axis is within ±10 degrees were left.Then the   was calculated using Equation 1, and   was calculated by dividing the instantaneous rotation of the wrist, Δ  , by the motion capture sample time  (   ≈ Δ  /).
Figure 6 shows a scatter plot of the ( [   ],  [   ]) points finally obtained from all participants.Notably, the mean mouse angular

W-index.
We calculated the W-index of each participant through the procedures in Section 3. We drew a W-diagram for each participant and performed linear regression independently (See Figure 8).The average  2 of the regression analysis was 0.998 (=0.0017).On average, W-index was 2.921, with the highest value of 3.463 and the lowest of 2.057.Table 2 summarizes the mean and standard deviation of the W-index for each game.Among the three genres, FPS was accompanied by the strongest wrist-aiming habits.
We also looked at the correlation between each participant's arm length and W-index, and the correlation between the number of clutches per minute and W-index (See Figure 7).However, none of them showed a significant correlation.In particular, we expected that a longer arm length would result in a longer arm rotation radius, resulting in a lower W-index, but the effect was insignificant.
The average arm postures of participants with the highest and lowest W-index are plotted in Figure 9. Overall, the participant with the lowest W-index tended to translate the arm-wrist-hand line in parallel.In the contour map, the participant with the highest W-index had a higher central density due to the rotational tendency of the wrist-hand line, but the participant with the lowest W-index showed a relatively uniform density area due to the parallel movement of the wrist-hand line.

STUDY 2: EFFECT ON WORKLOAD AND PERFORMANCE
Players believe that wrist-aiming habits have a significant impact on performance and upper-limit workload.In this study, we examine the relationship between W-index, input performance, and upper limb workload for two different groups of FPS players: professional and amateur.The workload was operationalized from the force the players exerted on the mouse, measured from a force sensor placed under the mouse pad.Considering the external validity of the experiment, we included a wide range of independent variables in the experimental design that could potentially affect wrist aiming habits (e.g., mouse sensitivity).In the first and third plots, gray lines refer to the shoulder line, green lines refer to the upper arm, blue lines refer to the forearm, and red lines refer to wrist to hand.In the second and fourth plots, darker red indicates the higher density.

Method
All professional players answered that they have been playing FPS games for more than 7 years, and amateur players answered an average of 4.45 years (=2.53).Unlike professional players, amateur players enjoyed a variety of FPS games: PUBG ( =4), VALORANT ( =3), Overwatch 2 ( =3), and Apex Legends ( =2).The mouse sensitivity settings normally used by professional participants were 800 CPI for four and 400 CPI for the other.Four of the amateur participants reported not knowing their sensitivity setting, while the remaining eight participants reported an average of 816.25 CPI (=235.1) as their usual setting.All participants were right-handed and had no upper limb injuries at the time of the experiment.

Task.
The most frequent and important task in first-person shooters is aim-and-shoot.In the task, a stationary or moving target appears on the screen.Players must rotate the first-person view camera through mouse control so that the crosshair in the center of the screen overlaps the target.Then, when the crosshair is positioned on the target, players must fire the weapon to eliminate the target6 (see Figure 10a).In this study, participants repeated the aim-and-shoot tasks using a dual-sensor mouse in a controlled desktop environment.The aim-and-shoot task given to participants in this study was implemented via FirstPersonScience (FPSci), a recently published library that can widely implement FPS-style user studies [5,28,31,51,52].FPSci provides various presets of target movements (see Figure 11) and allows precise recording of participant behavior within a task.The Player Group is a between-subject factor, and the other two are both within-subject factors.Among the pairs of words representing each level of the Task Format, the first means the type of movement of the target (Static, Straight, Stray), and the second means the type of weapon (Gun, Laser).Static refers to the fixed target, Straight refers to the target that linearly moves at a constant speed, and Stray refers to the target that moves in a straight line and changes its direction randomly in every random period between 0.75 and 0.9 seconds (Figure 11).Gun is a weapon that can end the aimand-shoot trial by pressing the mouse button only once when the crosshair overlaps the target, and Laser is a weapon that ends the trial when the crosshair overlaps the target for more than 1 second.
The game environment was rendered with a field of view of 103 • in width and 70 • in height.For each trial, the target's position and velocity were randomized.The target position was randomly initialized within the range 5 • to 15 • in azimuth and 0 • to 1 • in elevation to the reference direction (both azimuth and elevation are zero).The visual size of the target was fixed at 1.50 • (4 mm on monitor) for Gun condition and 2.55 • for Laser condition (7 mm).The speed of the moving target was randomized in the range 8 • /s to 15 • /s (approximately 3.1 cm/s to 5.7 cm/s) in the Gun condition and 10 • /s to 20 • /s (3.9 cm/s to 7.7 cm/s) in the Laser condition.These target conditions followed the settings of the FPSci-based experiment in [51].A summary is given in Table 3. Unlike Study 1, in Study 2, the mouse sensitivity was tested for a participant at various levels.In an FPS environment, mouse sensitivity means how much the first-person view camera rotates per millimeter movement of the mouse and has a unit of • /mm.Referring to previous studies, the range of sensitivity was determined to be between 0.5 • /mm and 2 • /mm as wide as possible [5].Players' preferred sensitivities usually fall within this range.Custom condition reproduces the sensitivity setting that each participant habitually uses.The condition was implemented in advance for each participant based on the results of the preliminary survey 7 .
In this study, we measure the following information while participants perform an aim-and-shoot task: (1) raw measurements from the dual-sensor mouse ( 1 ,  1 ,  2 ,  2 ), (2) 3D coordinates (, , ) of the joints of the participant's upper limb measured from the motion capture system, (3) raw pressure measurements from a force pad placed under the mouse pad (for workload analysis), ( 4) behavior logs recorded within FPSci (e.g., camera orientation, target trajectory, task execution time, task result).The sampling rates of the force pad and FPSci were 20 Hz and 240 Hz, respectively.
Professional players were asked the following three questions in a separate, short interview, and their answers were recorded: (1) Do you tend to aim with your wrist?, (2) How do you discuss your aiming habits with your colleagues?, (3) What aiming habits do you think reduce the chance of injury?5.1.4Procedure.Before participants arrived at the lab, we recalibrated the motion capture system.When participants arrived at the lab, they filled out a simple pre-survey and consent form and adjusted the desktop environment to be comfortable for them.The position of the mouse pad was also allowed to be changed freely by the participants, but the force pad was always placed under the mouse pad in the same way (Figure 10b).Then, a total of three optical markers were attached to the center of the hand, wrist joint, and elbow joint of the participants (Figure 10c).The experimenter asked the participant if the Custom sensitivity was properly implemented, and additional adjustments were made if necessary.Before entering the main part, participants practiced each level of Task Format for at least five trials under the Custom sensitivity setting.
For each participant, mouse sensitivity conditions were given in random order.Within one sensitivity level, the five Task Format levels were also given in random order.The next level starts only when the previous level is completed.The number of aim-and-shoot trials performed for each condition is shown in Table 3.In the Gun condition, the participants could hit the button repeatedly until they succeeded in shooting the target, but there was a 0.5-second cooldown time between button presses.If the trial in Gun condition and Laser condition was not completed within 5 seconds, it was considered a failure and proceeded to the next trial.720 trials per participant were conducted for about two hours.Participants were free to take breaks between sessions and could stop participating at any time if they wanted.Interviews with only professional players were conducted after the main experiment was completed and took up to 15 minutes for each participant.Participants were compensated with a gift worth 40 USD.

Apparatus.
The study was conducted using the same desktop computer, monitor, dual-sensor mouse, and motion capture system as in Study 1.As for the mouse pad, a general cloth surface type was used again.A mouse bungee was also provided for those who wished to use it (Razer Bungee V3).For the force pad, the Pressure Mat Dev Kit 2.0 model of Sensing Tex company was used, and it was confirmed in advance that the pressure measured with it showed a high correlation with the actual applied pressure ( 2 =0.99, see Figure 12).In subsequent analyses, measurements from the pad were converted to grams based on the regression equation.According to the official specification, the 32 cm × 32 cm measurement area is divided into 16×16 rectangles, and the applied force is returned for each rectangle.The mouse pad was completely contained within the measurement area of the pad with a slight margin (Figure 10).
Data from the mouse and motion capture system were collected using almost the same software as in Study 1. Force pad signals were received via a serial connection with a Python logger.All logging threads were triggered simultaneously with a single script.  1 Pre-processing.The collected mouse and motion capture data were pre-processed using the same process as in Section 4.2.1.The delay between force sensing pad data and mouse data was also estimated separately from cross-correlation (42 ms), and the timestamps of all data were unified.The force sensing pad was calibrated so that the sum of the forces measured when only the mouse and mouse pad were placed quietly was zero.

Descriptive Statistics.
In the Custom condition, participants' mouse sensitivity was 1.1 on average (=0.32).This is similar to what was reported in previous studies [5].Professional players' preference sensitivity was higher (=1.13,=0.28)than that of amateur players (=1.09,=0.36).The average arm lengths measured with a ruler for professional and amateur players were 55.6 cm (=4.2) and 55 cm (=3.4),respectively.Table 4 summarizes the number of clutches per minute, time spent per clutch, and average mouse speed (both longitudinal and lateral components) for each Player Group.The participants have indicated the mice they typically use as follows: Logitech G Pro X Superlight (63 grams, 3 professional, 1 amateur), Razer DeathAdder V3 Pro (63 grams, 1 professional), ZOWIE EC2-CW (77 grams, 1 professional), Zowie EC1-C (80 grams, 1 amateur), Logitech G102 PRODIGY (85 grams, 2 amateur), Logitech G302 DAEDALUS PRIME (87 grams, 2 amateur), ROCCAT Kone Pure Owl-Eye (88 grams, 1 amateur), Logitech G304 (99 grams, 2 amateur), Samsung AA-SM7PCP (100 grams, 1 amateur), Logitech G402 (108 grams, 1 amateur).One amateur participant was unable to specify the mouse model.The dual-sensor mouse used in the experiment had characteristics similar to those of the Logitech G Pro X Superlight in its shape and weight, which was the most favored mouse in our demographic survey.

Correlation between 𝐸 [ 𝜃 𝑤 ] and 𝐸 [ 𝜃 𝑚 ]
. Study 1 has already demonstrated that the mean angular velocity of the wrist  [   ] can be robustly estimated from dual-sensor mouse data.However, because each participant in Study 1 used their own mouse sensitivity setting, it was not tested whether the dual-sensor mouse technique was still valid over a sufficiently wide sensitivity range.In Study 2, each participant tested a wide range of mouse sensitivities, so we decided to confirm once again the correlation.
Binned averaging was performed in the same way as in Study 1. Figure 13

W-index.
We calculated the W-index for each of the 20 conditions for each participant (5 Task Format ×4 Mouse Sensitivity) (Figure 14).The method of calculating the W-index remained the same as in Study 1.A W-diagram was drawn for each condition, and linear regression was performed.The average  2 of the regression analysis was 0.996 (=0.004).The highest average W-index of 4.176 was observed from a professional player.The lowest average W-index of 2.248 was observed from an amateur player.With the table, we performed 2×5×4 mixed ANOVA with an  level of 0.05.When the sphericity assumption was not satisfied, the Greenhouse-Geisser correction was applied.Bonferroni correction was applied to the post-hoc analysis.
The main effect of Player Group on the W-index was statistically significant ( 1,15 =8.580, =0.010, 2  =0.364).The average W-index of the Amateur group or Professional group was 2.907 (=0.354)and 3.494 (=0.539),respectively.Professional players showed stronger wrist-aiming habits than amateurs.None of the interaction effects between Player Group and Task Format or Player Group and Mouse Sensitivity on W-index were significant (>0.197).
The main effect of Mouse Sensitivity on the W-index was also statistically significant ( 1.739,26.081=19.492, <0.001, 2  =0.565).The mean and standard deviation of the W-index for each sensitivity level are as follows: 0.5 (=2.883,=0.430),1.0 (=3.100,=0.506), 2 (=3.205,=0.452),Custom (=3.131,=0.534).As mouse sensitivity increased, wrist-aiming habits became stronger.Note that Custom sensitivity was 1.099 on average.The interaction effect between Mouse Sensitivity and Task Format on W-index was statistically significant ( 5.196,77.946=4.836, <0.001, 2  =0.244).When mouse sensitivity was as low as 0.5, participants tended to show a lower W-index in the Laser condition (Figure 15).The main effect of Task Format on the W-index was not statistically significant ( 2.642,39.631=1.470, =0.240, 2  =0.089).5.2.5 Workload.Workload was calculated as the average force applied to the force-sensing pad for each participant in each task condition.Kinematically, a higher force applied to the pad indirectly means that an overall stronger torque is acting on the limb joints.As stronger force is applied to the pad, the friction between the pad and the mouse increases, so greater torque is required to rotate the wrist to move the mouse laterally (i.e., radial-ulnar deviation).With the workload table, we performed 2×5×4 mixed ANOVA with an  level of 0.05.When the sphericity assumption was not satisfied, the Greenhouse-Geisser correction was applied.Bonferroni correction was applied to the post-hoc analysis.
The main effect of the Player Group on the mean force applied to the pad was statistically significant ( 1,15 =6.428, =0.023, 2   =0.300).Interestingly, the force exerted by professional players was higher (=2,687 grams, =199) than that of amateur players (=2,423, =245).None of the interaction effects between Player Group and Task Format or Player Group and Mouse Sensitivity on the mean force were significant (>0.688).

5.2.6
Performance.Player performance was quantified by two variables: (1) the average time taken for successful trials (Time), and (2) the percentage of trials missed due to failure to acquire the target within the time limit (Miss).We performed 2×5×4 mixed ANOVA with an  level of 0.05 for each of the two performance variables.When the sphericity assumption was not satisfied, the Greenhouse-Geisser correction was applied.Bonferroni correction was applied to the post-hoc analysis.
Mouse Sensitivity had a significant effect on Time ( 3,45 =5.903, =0.002, 2  =0.282) but not on Miss (=0.201).Players performed best at the intermediate level of sensitivity rather than at either extreme, replicating findings from previous studies [5] (see Figure 17).The interaction effect between Mouse Sensitivity and Task Format was not significant on both Time and Miss (>0.068).

5.2.7
Correlation between Dependent Variables.In this section, we look at the following two correlations between the dependent variables: (1) W-index vs. workload, (2) W-index vs. performance.Note that since the W-index was not controlled as an independent variable in this study, the correlations should not be misunderstood as causal relationships.
From the analysis, it was found that when the W-index increases, the force applied to the pad increases overall (Figure 18, left).We performed a linear regression of the W-index and force on the pad for each participant and obtained the average slope 295.7 (=169.1),and the average  2 , 0.213 (=0.148).Among all players, regression was significant for 9 players (<0.029).Professional players' slopes (=217.2,=124.9)were, on average, lower than those of amateurs (=328.4,=178.7).The relationship between W-index and performance showed no notable trends for both Time and Miss (see Figure 18).A linear regression was performed between performance and W-index for each participant, and the mean and standard deviation of slope and  2 are as follows: slope (=0.0896,=1.7532) and  2 (=0.0923,=0.0946) for Time, slope (=-0.3670,=8.1582) and  2 (=0.0581,=0.0857) for Miss (Figure 18, center and right).For Time, the correlation was significant for four participants (<0.043).For three of them, time decreased when the W-index increased.For Miss, the correlation was significant for two participants (<0.033).For both, Miss decreased as the W-index increased.

Interview Results
. Participants in Professional group subjectively evaluated their aiming habits as follows: strong wrist-aiming ( =2), medium wrist-aiming ( =1), weak wrist-aiming ( =2).Interestingly, for all but one participant, the measured W-index was consistent with their self-evaluation.Interestingly, with the exception of one participant, the measured W-index was consistent with their self-evaluation (Figure 19).
All participants responded they had experience discussing their aiming habits with other players.However, one participant said, "Because the intensity of the wrist-aiming habit is judged subjectively by the individual, the objective comparison is impossible, and therefore I think it is meaningless to discuss it with others", suggesting the need for standard operationalization of the wrist-aiming habit.Another participant said, "We usually discuss mouse sensitivity with each other.Players who use low sensitivity generally seem

W-index
Figure 19: Professional players' self-estimated wrist-aiming habit strength vs. their average W-index to have arm-aiming habits, and players who use high sensitivity seem to have wrist-aiming habits".One participant responded, "In official competitions, I sometimes use low mouse sensitivity because I get nervous, and my hands can shake." Except for one participant who answered that aiming habits had no effect on wrist injuries, the remaining four participants answered that having arm-aiming habits would be more helpful in preventing injuries.

DISCUSSION AND IMPLICATIONS
We found that the lower the Mouse Sensitivity, the more likely players were to aim with their arms (i.e., lower W-index).This seems to be a natural effect because lower Mouse Sensitivity requires moving the mouse over a wider area that is difficult to cover with just wrist rotation.This is an effect that is already well-known among players on the web.One interesting finding is that as Mouse Sensitivity increased, even though players were free to choose between wrist-aiming or arm-aiming, their behavioral policies shifted toward wrist-aiming (see Figure 15).The size of this effect was notable ( 2  =0.565), leading us to speculate that wrist-aiming may go beyond individual tastes and habits and provide fundamentally higher benefits in the aim-and-shoot task covered in Study 2. The benefits may not simply be related to higher performance, but also lower energy consumption [16], since rotating only the wrist will consume less energy than rotating the entire arm.In fact, in Study 2, professional players showed stronger wrist-aiming habits than amateur players, but the positive correlation between W-index and performance was weak only for some players.
We note that a previous study of another FPS game (Counter-Strike: Global Offensive or CS:GO) found that professional players primarily arm-aimed with lower mouse sensitivity than amateur players [49].CS:GO is classified as a more traditional FPS game than Rainbow Six game played by professional players in this study), where pure aiming ability is more important than special moves or item use.Similarly, a significant effect of Task Format on W-index was observed in Study 2 (see Laser condition in Figure 15).From these results we cautiously speculate: a higher W-index is partially associated with higher performance, but even within the same FPS genre, the appropriate strength of wrist-aiming habits seems to vary depending on the mechanics [18] required within the game and the goals players are pursuing.In that regard, we also hope that this study will trigger future modeling studies on the hidden mechanisms by which a player's wrist-aiming habits are rationally determined [48]; Recent modeling studies of pointing behavior [9,16,22,24] have just begun to consider the kinematics and dynamics of the upper limb.
As Mouse Sensitivity increased, participants showed stronger wrist-aiming habits, which was associated with higher upper limb workload (i.e., stronger pressing force).However, in this study, we did not ask participants to change their wrist-aiming habits while controlling for Mouse Sensitivity.Therefore, it is difficult for us to determine whether the increase in workload is due to an increase in Mouse Sensitivity or an increase in W-index.However, it is noteworthy that in Study 2, the W-index and mean force (workload) measured in 20 different conditions for each participant showed a consistent positive correlation (Figure 18).Therefore, in this study, we give weight to the interpretation that stronger wrist-aiming places a greater workload on the upper limb.
Then why do players apply more force when wrist-aiming?Wrist-aiming is considered a process of controlling a smaller mass than arm-aiming (hand vs. arm), so although movements may be faster, stable control may be more difficult for players.Applying more force to the mouse increases the friction between the mouse and the desk surface and also increases the stiffness of the muscles [14], possibly allowing for more stable control.In a professional player's opinion that he uses low mouse sensitivity if he has severe hand tremors due to nervousness, we can also get a hint that wristaiming habits are deeply related to stability issues.
As evidenced by interviews with professional players, our technique will provide a standard, objective measure of wrist-aiming habits, making it possible for players to discuss and compare their habits with each other.From there, players will be able to discover better personalized settings and training routines.For example, in the interview, a player with a relatively low W-index (2.817) said he thought he had strong wrist-aiming habits, which may have caused him to train with more concern about wrist injuries than necessary.Furthermore, the high accessibility of the technique will allow the community to accumulate large-scale data related to wristaiming habits, which can finally provide answers to long-standing problems.For example, let's say that for hundreds of professional players, their W-index was tracked over a relatively long period of time.If some of those players retired early due to wrist injuries, analyzing their W-index could help answer the question of whether stronger wrist-aiming habits actually increase the likelihood of wrist injuries.If discussions on W-index become more active, there is a possibility that dual-sensor mice will be mass-produced by leading manufacturers (e.g., Logitech or Razer), since it is a matter of adding one more sensor (not very expensive) to an existing mouse.

LIMITATIONS
Despite its methodological and empirical contributions, this study has several obvious limitations.First, the W-index presented in this study is only one of the various operationalization possibilities of wrist-aiming habits.In particular, the W-index only tells us the average strength of the wrist-aiming habit, but other indices, such as the maximum and minimum deviation of the wrist joint or the frequency of changes in the direction of wrist rotation, may provide additional information that the W-index does not take into account.
Second, the high correlation between the average angular velocity of the mouse and the average angular velocity of the wrist was discovered by chance, and this study does not explain the underlying mechanism.
Third, our study addressed the relationship between W-index and player performance only through controlled tasks.Additional research is needed on the impact of wrist-aiming habits during free gameplay, for example in direct confrontations with enemies or when ammunition is a finite resource.
Fourth, the assumption that the center of the mouse overlaps the center of the hand may not always hold in reality; It is known that players use a variety of methods to hold their mouse [43].If the offset between the mouse center and the hand center is unexpectedly large, lateral mouse speed estimates may be affected [30], which may lead to unintended differences in W-index.W-index compensation according to mouse grip is also a good topic for future research.
Fifth, in Study 1, the order in which games were given to participants was not counterbalanced, so the carry-over effect may not have been sufficiently eliminated.Exploring the impact of learning and fatigue on the W-index is also a potential future research topic.
Sixth, in this study, the players' upper limb workload was indirectly inferred from the mean force applied to the pad.However, the force applied to the pad includes the original weight of the arm 8 , and if part of the arm goes outside the pad and is pressing the desk, that partial force cannot be measured.Attempting more direct measurements, such as electromyography sensors, is a potential topic for future research.Longitudinal studies of the relationship between W-index and upper limb injuries also need to be conducted.
Lastly, we operationalized player performance into just two aggregate statistics: Time and Miss.However, kinematic metrics that can evaluate the quality of movement trajectories, such as path efficiency [54], movement variability [41], and number of submovements [32,34], should also be addressed in future research.

CONCLUSION
The computer mouse has become the primary input device used by millions of players in desktop competitive video games today (i.e., esports).In desktop video games, wrist-aiming habit refers to the degree to which a player relies on wrist rotation when moving a computer mouse.Despite high demand, there has been no scientific investigation into how wrist-aiming habits affect gaming performance and upper limb workload.In particular, the lack of affordable techniques to quantify wrist-aiming habits has been a major obstacle to scientific investigation.
In this study, we presented a novel technique to quantify the extent of wrist-aiming habits of computer mouse users through an affordable device called a dual-sensor mouse.The technique calculates the W-index, a scalar value proportional to the average strength of a user's wrist aiming habits, from only the dual-sensor

Figure 3 :
Figure 3: Kinematics of mouse control: (a) The player arm can be modeled as a seven-degree-of-freedom (7 DOF) linkage in 3D.(b) Looking down at the desk from above, the arm can be simplified to a 3 DOF linkage.(c) The 3 DOF linkage has infinitely many solutions for determining the mouse position (  ,   ) on the desk (i.e., a redundant system).

Figure 4 :
Figure 4: A dual-sensor mouse used to estimate the average angular velocity of a player's wrist joint ( [   ]).

Figure 5 :
Figure 5: (a) Motion capture settings in Study 1 and 2, (b-c) front and back of the dual-sensor mouse used in the experiment, (d) location of markers attached in Study 1.

Figure 6 :
Figure 6: Correlation between mean mouse angular velocity measured from a dual-sensor mouse and mean wrist angular velocity measured through motion capture.Note that all data points of the 20 participants are plotted without aggregation.

Figure 7 :
Figure 7: (Above) Relationship between participant's arm length and mean W-index, (Below) Relationship between participants' clutching frequency and W-index.

Figure 8 :Figure 9 :
Figure 8: W-diagram and  2 values between  [  ] and  [   ] for each participant and three different games

Figure 10 :
Figure 10: (a) Screen of the aim-and-shoot task given in Study 2, (b) Force-sensing pad used (located under the mouse pad), (c) Attachment location of motion capture marker in Study 2

Figure 11 :
Figure 11: Target movement conditions given in Study 2

Figure 12 :
Figure 12: The averaged raw signal measured from the force sensing pad (-axis) vs. the actual applied force (-axis) shows the scatter plot of  [   ] and  [   ] for all participants and all sensitivities.The corresponding linear regression equation is ( 2 =0.9473):  [   ] = 0.7085 •  [   ] − 0.000710.This again demonstrates the feasibility of quantifying wrist-aiming habits using only signals from a dual-sensor mouse, over a wide mouse sensitivity range.The regression equations are also almost similar (Study 1:  [   ] = 0.7039 •  [   ] + 0.000175).In the subsequent W-index calculation, the following equation was used by averaging the two regression equations:  [   ] = 0.7062 •  [   ] − 0.0002675.

Figure 13 :
Figure 13: Relationship between mean mouse angular velocity and mean wrist angular velocity for all participants in Study 2 for different mouse sensitivities.

Figure 17 :
Figure17: Significant effects of each independent variable and their interactions on the mean time, which is an aim-and-shoot performance variable; The error bar represents a 95% confidence interval.Statistically significant differences are indicated ( * :  < 0.05, * * :  < 0.01).The post-hoc analysis is reported for the main effects only.

Figure 18 :
Figure 18: Study 2: Relationship between W-index, mean force applied to the pad, mean completion time, and miss, obtained for each condition and each participant (20 points per participant).

Table 1 :
Descriptive statistics of participant behavior in Study 1 calculated after pre-processing   ] and the mean wrist angular velocity  [   ] show a correlation with a significantly high coefficient of determination ( 2 =0.98).This demonstrates the feasibility of quantifying wristaiming habits using only signals from a dual-sensor mouse.The linear regression equation was:  [   ] = 0.7039 •  [   ] + 0.000175.Note that this single regression equation accounts for all participants' data.As we expected,  [   ] was smaller than  [   ] because the rotation of other joints also contributed to  [   ].

Table 2 :
and  of W-index for each game genre

Table 3 :
Implementation of five Task Format conditions

Table 4 :
Descriptive statistics of participant behavior in Study 2 calculated after pre-processing; Note that clutching statistics are calculated excluding conditions where clutching is not observed.