The player experience is a vital part of a game. The quality of the interactions between the game and the player is often seen as the definition of player experience, eventually resulting in the player's enjoyment. For my master thesis, I tried to discover if the player experience could be enhanced by using real-time biometric data to adapt the mood of a virtual reality game environment.
Implementing biometric measurements in games can be done in multiple ways. For instance, by studying the social experience, game features, game events, and game design improvement. Alternatively, real-time biometric data could be used to manipulate in-game data, such as controlling a character or player functions. This is described in more detail in my previous article.
For this master thesis, a VR corn maze game was created. It had two different versions, one where the mood and weather of the game were manipulated based on the real-time biometric data of the participant. The other version was a set experience without mood changes. In the first version, the mood and weather got worse based on the arousal level of the participant; this happened when the arousal level was above a set threshold. The biometric measurements were done with the Empatica E4 wristband; the arousal was measured through skin conductance from the participant's hand. My previous article goes into more detail about the Python plug-in and the VR corn maze. This article will mainly focus on the data collection, data analysis, discussion, recommendation, and the future indications of this study. The complete thesis can be found at the end of this article.
The data collection
Thirty-four participants were recruited, which due to COVID-19 was slightly less than the desired 40 participants. A convenience sample was used because it was difficult to get participants during the COVID-19 pandemic. The participants were randomly and evenly assigned to one of two groups, an adapted experience group or a controlled experience group. The adapted experience group navigated through a version where the mood and weather got worse when their arousal increased and got better when their arousal declined. Whereas the controlled experience group had a version wherein the mood and weather stayed constant.
Calm mood Heavy storm
Before the experience, a questionnaire was used to measure the participants' emotions, arousal, and valence, using the Scale of Positive and Negative Experience. After this, the participant could play the experience. The biometric data used during the experience was not collected due to the scope of the project. After the experience, a post-experience questionnaire should be filled in. This questionnaire included the previously filled-in pre-experience questionnaire at the beginning. Then the competence, the enjoyment of the art style, immersion, and the experience with VR were also measured. The questionnaire ended with the intent to recommend the experience. The post-experience questionnaire used a mix of the Game Experience Questionnaire and the Player Experience of Need and Satisfaction.
Pre-experience questionnaire Post-experience questionnaire
The data analysis
Out of the 34 participants, 20 participants were female. Leaving 14 male participants. The ages of the participants ranged from 12 to 69, with an average age of 27.5. Due to this research taking place in the Netherlands, 28 participants were Dutch. There also were four Bulgarian participants, one German participant, and one Zimbabwean participant. The minority, 12 participants, had experienced VR in the previous 12 months. Ten of these 12 participants had experienced VR once; the other two participants used VR between 21-30 and 50+ times. Of the 12 participants that had experienced VR, ten participants have experience with interactive VR; the other two participants only experienced the sitting type of VR. The average time the experience took was a little over eight minutes. If wrong paths were taken, the participants encounter dead ends, leading to a longer completion time.
Genders in the research groups Nationalities of the participants
Overall the participants felt 6.3% better after the experience than before. After the experience, the adapted experience group felt 3.7% less good than before the experience. Whereas the controlled experience group felt 15.4% better. There is a negative correlation between the player's emotions after the experience and the research group, with a correlation coefficient of -0.310 and a P-value of 0.034. The majority, 58,9% of the participants, thought the experience was not hard. However, 18 participants felt challenged. These participants took the wrong path multiple times; they noted that they felt challenged by the maze, but they did not find the experience hard. Overall the art style was appreciated by the participants. Most participants, 88.2%, felt completely absorbed, 73.5% felt they returned from a journey, and 73.5% felt they lost track of time. The adapted experience group felt 3.6% more absorbed. However, they did feel 37.1% more distracted and 21.8% more disengaged. They noted that they felt distracted by the rain. All participants think VR added immersion to the experience. However, six participants felt slightly nauseous during the experience, and two participants felt slightly nauseous after the experience. This was caused mainly by standing too close to the sensors, causing a white flash.
All means per research groups
Regarding the likelihood to recommend the experience to a friend or family, the participants gave an average grade of 8.18. Female participants tended to give a slightly higher grade than the male participants. The adapted experience group gave a slightly lower grade than the controlled experience group. The overall grade of the entire experience was slightly lower than the likelihood to recommend the experience, an 8.12. For this question, female participants also rated the experience slightly higher than the male participants. The average grade of the adapted experience group was slightly lower than the controlled experience group.
Intent to recommend Overall grade of the experience
There was a small but noticeable difference between the experience of the adapted experience group and the controlled experience group. The adapted experience group had a lower overall grade for all post-experience questionnaire questions. This means that they rated their experience to be less than the controlled experience group. Overall the emotions and feelings of the participants became more positive. However, the adapted experience group was less positive after the experience than the controlled experience group. Multiple participants have noted that the rain made the experience harder, and they felt less competent. The adapted experience group also rated the art style less pleasant; however, multiple participants noted that they thought the environment looked "cooler" in the dark and stormy mood. The experience was also less immersive for the adapted experience group. The VR experience was rated less pleasant for the adapted experience group.
Most likely, the lower ratings, thus the lesser experience, were due to the adapted experience group getting a negative feedback loop from their biometric data and the negative connotations with rain and storm. When the participant had high arousal levels, they would get a slowly rising storm. High biometric measurements meant a heavy storm; this negatively impacted the emotions and experience of the participant. The storm also made it harder to see and a slightly less pleasant experience for the eyes, explaining why participants from the adapted experience group had a harder time, rated the art style less pleasant, and rated the VR experience less pleasant. There is an indication that the player experience can be impacted by using live biometric data; in this case, mood and weather changes. The indication of this opens up the possibility of improving the player experience by using a positive feedback loop instead.
Recommendations and future indications
For this specific research, I would suggest further investigation to fully understand what type of impact the dynamic mood-changing has on the player experience. The current data suggest a connection between the two, which in this case is a negative connection. I assume a positive connection could be possible when the mood changing would be implemented differently; this would need further investigation. In order to create a more accurate assessment of the research, I would advise increasing the number of participants. It would also be interesting to see the data of a more diverse sample, for instance, different nationalities. Since the research was conducted in the Netherlands using a convenience sample, the sample was mainly western European, young adults, and inexperienced VR users.
Quite a lot of research is done on using biometrics in GUR; it is one of the biggest biometrics implementations in the gaming industry. However, this specific framework does not fit this implementation well. Unless the GUR research is about a related topic to this research, such as "What type of game fits game modification by real-time biometric data best?".
Improving the player experience using real-time biometric data could be taken a step further and create a personalized experience, such as personalized music, character features, or environmental features; however, this needs further research.
It would be interesting to see if a dynamic difficulty system based on real-time biometric data would have an impact on the player experience. I suspect it could improve the enjoyment of the player because they might not give up as easily. Further research should include a way to reverse the difficulty system as some gamers enjoy challenging games. It could also improve serious games when it adjusts to the player's level.
I have had many discussions with fellow game artists about using biometric measurements to increase psychological horror elements in games. Here, a threshold could be set that only if the participant passes the threshold makes a horror event happen. If this threshold is not reached, the tension in the game could be increased to get the biometric data to cross that threshold.
Thank you for taking the time to read through my article, and I hope you find it just as interesting as I do! Feel free to read the complete research.