Competition and cooperation with virtual players in an exergame

Two cross-sectional studies investigated the effects of competition and cooperationwith virtual players on exercise performance in an immersive virtual reality (VR) cycle exergame. Study 1 examined the effects of: (1) self-competition whereby participants playedtheexergame while competing against a replayof theirpreviousexergame session (Ghost condition), and (2) playing the exergame with a virtual trainer present (Trainer condition) on distance travelled and calories expended while cycling. Study 2 examined the effects of (1) competition with a virtual trainer system (Competitive condition) and (2) cooperation with a virtual trainer system (Cooperative condition). Post exergame enjoyment and motivation were also assessed. The results of Study 1 showed that the trainer system elicited a lesser distance travelled than when playing with a ghost or on one’s own. These results also showed that competing against a ghost was more enjoyable than playing on one’s own or with the virtual trainer. There was no signiﬁcant difference between the participants’ rated enjoyment and motivation and their distance travelled or calories burned. The ﬁndings of Study 2 showed that the competitive trainer elicited a greater distance travelled and caloric expenditure, and was rated as more motivating. As in Study 1, enjoyment and motivation were not correlated with distance travelled and calories burned. Conclusion: Taken together, these results demonstrate that a competitive experience in exergaming is an effective tool to elicit higher levels of exercise from the user, and can be achieved through virtual substitutes for another human player.


INTRODUCTION
competitive or cooperative experience. The first is a "ghost-replay" system, in which the player is able 48 to record play sessions and then compete against either their own recordings or the recordings of other 49 players. In such a replay system, the user should always be motivated to improve, by focusing on beating 50 the ghost of their last attempt. The second is an AI player in the form of a virtual "trainer" system, which 51 adapts to the fitness level of the user. We present two user studies. The first compares the ghost replay 52 system with a simple AI trainer. The second utilities a more advanced trainer system, based on the design 53 of the first training system but allowing for differing behaviour profiles. This second study compares two 54 trainer profiles: one that competes with the player and one that cooperates with them. 55 Using these two studies, this paper aims to answer the following research questions: 56 R1 How does self-competition provided by a ghost replay system influence the user's enjoyment, 57 motivation, or exercise performance during play of a virtual reality exergame? 58 R2 How does playing with a competitive or cooperative trainer system influence the user's enjoyment, 59 motivation, or exercise performance during play of a virtual reality exergame? 60 R3 How does the competitive inclination of the user influence the effectiveness of a competitive or 61 cooperative trainer system on the user? 62 Based on existing research in this area discussed in the next section, we hypothesise that self competi-63 tion via the ghost replay system should increase both the user's enjoyment and motivation in the exergame, 64 as well as their overall exercise performance. We hypothesise less of an effect for a trainer system than 65 the ghost replay system, but expect that a trainer system will be more effective when aligned with the 66 user's personality. There is evidence that the use of Virtual Training systems (Virtual Trainers) influences users' motiva-97 tion and exercise adherence, and may avoid some of the downsides of traditional multiplayer gaming. In 98 particular, situations in which an individual feels stigmatised can affect exercise motivation negatively by 99 increasing anxiety and avoidant behaviours (Lantz et al., 1997). 100 Current research on virtual trainers has focused on the use of a trainer separated from the gameplay. 101 Ijsselsteijn et al. (2006) studied an exergame in which a virtual coach provided users with regular feedback 102 about their heart rate. The trainer was a virtual human female character that was displayed in the corner 103 of the screen. The feedback was provided in the form of pre-recorded voice cues and corresponding text 104 shown in a speech bubble above the coach, e.g. "Your heart rate is too low. Cycle faster." The trainer 105 lowered tension surrounding performance and player control, while not affecting enjoyment. The results 106 also indicated that greater immersion in the game is linked with increased motivation.

107
The direct instructions used in the aforementioned study by Ijsselsteijn et al. (2006) have potential 108 downsides. Hepler et al. (2012) report that the effectiveness of these prompts and cues may rely on the 109 personality and past behaviour of the user. For example, a user with a history of sedentary behaviour may 110 ignore an instruction such as "cycle faster". Furthermore, the user's interpretation of feedback can have 111 a significant effect on how it motivates the user. If the feedback is interpreted as controlling, the user 112 may not be inclined to respond to it (Deci and Ryan, 1985). As a consequence, cues to exercise harder 113 when the current level of exertion is insufficient should not be presented in a way that may be perceived 114 as controlling, as this is likely detrimental to motivation. system, the user's bodily motion was detected using a Kinect 3D sensor, and the user gained points 117 by mimicking the motions shown on screen by the trainer. While this system had a limited degree of 118 gamification, the research indicates that training in an immersive virtual environment is motivating. 119 Wilson and Brooks (2013) compared training with a virtual trainer in an exergame to training with a 120 certified human trainer in a traditional exercise program. While the levels of exertion (measured by heart 121 rate and rate of perceived exertion) are higher with a human trainer, the results showed no significant 122 difference in exercise adherence between the two trainer types.

123
In a similar study, Feltz et al. (2014) had participants completing exercises either alone, partnered 124 with a human, partnered with a human-like virtual player, or partnered with a non-human-like virtual 125 player. The partners were designed to appear more capable than the participant at the exercise task. In 126 similar results to Wilson and Brooks, exercise performance was higher with the human partner than the 127 virtual partners, but all partnered conditions showed higher performance than the solitary condition.

128
These two studies suggest that a properly designed virtual trainer could be suitable as a longer-term 129 motivational tool for exercise. Such a trainer would also likely improve health outcomes by encouraging a 130 greater degree of exercise performance. 131 While there has been a moderate amount of research on competition and cooperation in exergames, 132 this research has been heavily focused on the use of these factors with other human players. Similarly, 133 while there has been a moderate amount of research on virtual trainers, the training systems in existing 134 research have little gamification and do not look at competition or cooperation.

137
A cross-sectional within-subjects study was conducted to examine the effects of competition and coop-   For this research, we extended an existing exergame described in Shaw et al. (2015). This exergame was 157 chosen as it elicited high intensity exercise from the users, and was rated by the users as enjoyable. In 158 this exergame, the user cycles along a procedurally generated track, avoiding obstacles and collecting 159 bonuses, in an effort to maximise their score. The speed at which the user moves is governed by the rate 160 at which they pedal on the exercycle. A 3D camera tracks their movements, allowing them to steer by 161 leaning from side to side. The game is presented to the user via an Oculus Rift Head Mounted Display 162 (HMD), providing them with an immersive experience. 163 We extended this exergame, adding a replay system and a simple virtual trainer system. The base 164 gameplay was also slightly modified, changing obstacles to slow the player and penalize their score, rather 165 than causing them to replay a section. This was necessary in order to avoid divergence between the ghost 166 replay system described below, and the user's current play session.
167 Figure 1 shows a screenshot of the exergame, and illustrates some of the gameplay elements.

168
The exergame allows for playback of a participant's previous attempts through a "ghost racer" system, 169 in which the participant sees a non-interactive replay of the past attempt on the track as they play. This 170 offers encouragement to exercise harder in order to beat their previous attempt. When users are lagging 171 behind their "ghost", they are also able to see points where their previous run failed to avoid obstacles, 172 and thus they may be able to adapt their play to avoid more obstacles.

173
The player's ghost has the same appearance as the trainers (shown in figure 2): a simplified figure on 174 a bike. When close to the player, the ghost and trainers are semi-transparent, increasing in opacity as they 175 move further away. This is to prevent them from blocking the players view of obstacles or bonuses and 176 becoming a potential source of frustration.

177
The simple virtual trainer system behaves in a similar fashion to the ghost system. When the player 178 begins to move, another 'player' appears and travels along the track with them. However, rather than 179 showing previous behaviour, the trainer attempts to show optimal behaviour, both in terms of gameplay 180 and exercise. With regard to gameplay, the trainer chooses an optimal path through the track, avoiding all 181 obstacles. With regard to exercise, the trainer adjusts its speed to guide the user towards an ideal exercise 182 heart rate, as explained below.

Computer Science
First, the current heart rate of the user, as measured by the handlebar sensors, is used to estimate a 184 relative heart rate, that is, a percentage of the user's expected maximum heart rate based on their age. attempts to set a speed suitable for keeping the user's heart rate at the level associated with moderate to 187 vigorous exertion, that is, 64% -90% of their expected maximum heart rate (Garber et al., 2011). If the 188 user's heart rate is below 64% ("low heart rate"), the trainer increases its speed, requiring the user to work 189 harder to catch up. If their heart rate exceeds 90% of their maximum ("high heart rate"), it decreases its 190 speed, allowing them to exert less effort to keep pace. While the user's heart rate is in the target zone 191 ("average heart rate"), the trainer stays a short distance in front of the user providing a target to follow in 192 order to motivate them.  Participants first completed the ten-minute "Control" condition, followed by the "Ghost" and "Trainer"  (Borg, 1982), and asked to rate their level of exertion. They were also given a post-condition 204 questionnaire in which they were asked to rate how enjoyable and motivating they found the condition, 205 and were invited to give feedback about the Ghost and Trainer systems, and about the exergame in general.  The total kilocalories expended on the exercycle as the total Calories expended at the end of each exercise 212 session. This was measured from the exercycle's output. 214 The RPE scale (Borg, 1982) is a brief self-administered rating scale that was designed to measure an 215 individual's subjective rating of exercise intensity. At the end of each exercise session, participants rated 216 their perception of effort or "how hard they felt they had worked" during each exercise session, using a 217 scale ranging from "6" (least exertion) through to "20" (most exertion). Ghost, and Trainer conditions on distances travelled, calories expended, and rate of perceived exertion.

234
Due to the non-normally distributed data, the effects of the three conditions on enjoyment and 235 motivation were examined with Friedman tests.

236
Pearson correlation analyses were used to examine the association between the participant information 237 gathered during the pre-test, and the measures listed above.

Results and Discussion
239 Table 1 shows the means and standard deviations of the various measures across the three conditions.

240
The results of a RM-ANOVA showed that there is a significant main effect Condition on distance Trainer conditions.

253
The results of a Friedman test showed no significant main effect Condition on motivation across the 254 three conditions (p = .370).

255
The results of a Pearson correlation showed that enjoyment of a condition, and level of motivation in 256 that condition have no significant correlation with distance travelled in the condition.

257
The use of player recordings of past performance to encourage self-competition shows significant 258 promise to encourage users to exercise via an exergame, particularly if they enjoy competition. Verbal 259 and qualitative feedback from the participants indicated that being able to see and beat their previous 260 attempt was highly enjoyable during the Ghost condition. This study failed to show benefits for the use of 261 a multiplayer-style virtual trainer system, however that may be due to flaws in the trainer system discussed 262 further below.

263
It is not too surprising to see that the Ghost condition did not encourage players to exercise significantly citing unrealistic behaviour: "The trainer system moved strange". We suspect this may be related to 271 the framing of the trainer system. While the ghost system was clearly competitive, the trainer system 272 had no particular framing as either competitive or cooperative. If the trainer was ahead, it would show 273 participants an optimal performance, but participants were able to push themselves above the target heart 274 rate and pull ahead; "competing" with it.

275
The trainer system was extremely effective at avoiding obstacles, and often navigated through obstacles 276 with superhuman dexterity. When this occurred, participants tended to react negatively, stating that they 277 felt that the trainer was "cheating", and was not helping them as it was not showing them an optimal path 278 that they were capable of following.

279
It should however be noted that while the Ghost condition was better received than both the Control 280 and Trainer conditions, the overall participant response to all three of the conditions was generally positive, 281 with the mean enjoyability and motivation ratings still being high. The exergame in general was regarded 282 as enjoyable and motivating, and the Trainer system did not detract from that.

285
A cross-sectional within-subjects study was conducted to examine the effects of competition and co-

301
As research discussed earlier in this paper indicates, competition as part of an exergame can affect 302 different users very differently depending on how competitive they are. The behaviour of the first trainer 303 system was not clearly framed as either competitive or cooperative. The advanced trainer system was 304 designed to be customizable for either competition or cooperation in order to appeal to different personality 305 types. In order to do that, the advanced trainer implements two behaviour profiles: a competitive profile 306 and a cooperative one. While the competitive trainer profile is programmed to challenge and race against 307 the player, the cooperative trainer profile attempts to help the player achieve a higher score.

308
Similar to the previous trainer, the advanced trainer always chooses a path that is close to optimal for 309 scoring points and attempts to avoid obstacles. The trainer looks ahead to avoid obstacles in the distance.  Figures 1 and 2)  The advanced trainer modifies its behaviour based on the heart rate of the user, considering the same low, 322 average, and high heart rate zones as the simple trainer (see Figure 3). For the competitive trainer profile, 323 when the player's heart rate is too low, the trainer's speed will increase up to 1.3 times that of the player.

324
When in the average heart rate zone, the trainer's speed will approximately match that of the player. And 325 when the player is in the high heart rate zone, the trainer's speed will drop down to 0.7 times that of the 326 player.

327
The speed of the trainer also takes into consideration the distance from the player. If the player spends clamping the trainer's distance means that the player is always able to look over their shoulder and see the 334 trainer following them.

335
While in the target zone, the speed variation means that the trainer behaves as a human player of 336 similar abilities, in that it occasionally pulls slightly ahead and occasionally falls slightly behind. This

Manuscript to be reviewed
Computer Science similar fashion to the competitive trainer, but within the bounds given by its positions when the player is 353 in the high or low heart rate zones. (see Figure 4).

355
Participants completed a pre-test questionnaire to provide general demographic data: their age, gender, 356 and baseline self-report measures of the typical number of hours spent exercising and playing video 357 games each week. As part of the pre-experiment questionnaire, participants also filled out the Sport

358
Orientation Questionnaire (SOQ) (Gill and Deeter, 1988) and the Task and Ego Orientation in Sport 359 Questionnaire (TEOSQ) (Duda, 1989). These are validated and commonly used questionnaires that 360 provide five personality metrics related to competitiveness in sporting activities: competitiveness, goal 361 orientation, and winning orientation from the SOQ, and task and ego orientation from the TEOSQ.

362
Following the questionnaires, participants were then given a written outline of the test procedure and For each condition, the distance travelled in kilometers on the exercycle was assessed as the total 374 kilometers travelled at the end of each exercise session. This was measured from the exercycle's output.

376
The total kilocalories expended on the exercycle as the total Calories expended at the end of each exercise 377 session. This was measured from the exercycle's output.  significantly higher than in the Control condition (p = .022), and the Cooperative condition (p = .011).

407
There was no significant difference between the Control and Cooperative conditions.

408
The results of a Friedman test did not show a significant difference in enjoyment across the three 409 conditions (p = .756).

410
The results of a Friedman test showed a significant difference in motivation across the three conditions 411 (p = .027). The Competitive condition was significantly more motivating than the Control and Cooperative  These two studies suffer from some limitations in their experimental design and procedure. In Study 1, 466 the need for a dataset to be used by the ghost replay system meant that the default condition could not 467 be counterbalanced with the other two conditions. Additionally, as mentioned above if a participant was 468 beating their ghost, they would have to look behind in order to see it and compare their performance.

469
However, if they pulled too far ahead the ghost could end up too distant to see.

470
In Study 2, the user's rate of perceived exertion was not measured. As such, we are unable to see how  In both studies, the "enjoyment" and "motivation" constructs were only measured with a single 488 item in the post-condition questionnaires. This reduces their reliability in assessing the opinions of the 489 participants.

491
We have presented a set of systems for an immersive VR exergame that attempt to provide the benefits of 492 a multiplayer experience with regard to the use of competition and cooperation as a motivational tool.

493
Our results indicate that competition is a useful tool in exergaming, but do not show that that is 494 necessarily the case for cooperation. Virtual players, either a replay or an AI trainer provide an effective

Manuscript to be reviewed
Computer Science substitute for a human player in order to increase the motivation of the user, and can increase the user's 496 exercise performance. However a cooperative virtual player appears no more effective than solitary play.

497
Interestingly, our results do not indicate an influence for the personality of the player on what kind of 498 virtual trainer system they prefer.

499
Using the user's heart rate as a tool for governing the behaviour of a virtual trainer appears an effective 500 means of balancing the trainer's performance such that the user exercises at a worthwhile intensity.

501
There are two main implications that our results hold for the design of virtual players for use in 502 exergaming systems. Firstly, our studies indicate that a more interactive experience leads to greater 503 exercise intensity, likely through greater player investment in the experience. Secondly, the experience 504 should be clearly competitive or cooperative as an experience with unclear orientation may be less effective 505 than solitary play.