PORTFOLIO
Game & Psychology & Behavior
My Role: Research, Art, Qualitative Design, Experiment, Data Analysis
Duration: Jan.2022-Mar.2022
This is a two-stage playable and immersive psychological experiment, aiming to use games as a media to verify the existence of two psychological phenomena, including the "Broken window Theory" and "Theory of Mind".
Project Motivation and Inspiration
As people's acceptance and attention towards virtual and immersive experiences gradually increase, as developers deeply interested in human-computer interaction experiences, we attempt to explore human nature and behavior in the virtual world through experiments. By selecting two widely recognized psychological theories, we propose to integrate them into the gaming experience to test the presence of these theories in the virtual world. Unlike traditional testing, we hope to design experiments in a more engaging way to understand the participants' real psychological and decision-making behaviors.
Phase 1
1. "Broken Window Theory"
Broken windows theory is a popular theory in criminology that contends the state of an area can influence the decisions of individuals to commit crimes in that area. For example, an unkept and disordered area is more likely to encourage criminal activity due to the perception that crime in this area is unchecked. On the other hand, an ordered and clean area is less likely to encourage criminal activity due to the perception that crime in this area is heavily discouraged. Our research seeks to isolate this variable - the orderliness (or disorderliness) of an area - and measure its influence on player behavior, specifically in regards to committing non-violent crimes to achieve an in-game goal.
2. Experiment Design
Using the Unity engine, we have designed a city scavenger hunt game. We plan to achieve experimental purposes through qualitative survey and in-game data monitoring, therefore conducting our experiments through playtesting with our participants who are not informed with our testing subjects and goals.
2.1 Game Design
The game is crafted as a first-person 3D walking simulator where players must collect all the required items from a given list within a limited time frame.
Drawing inspiration from the Broken Window Theory, we emphasize the environment in two contrasting settings:
- Ordered Zone: A picturesque affluent street with no litter, visible security cameras, and a strong security presence.
- Disordered Zone: A dirty street with litter strewn about, a run-down appearance, and various obstacles along the way.
- Transitional Zones: To enhance environmental realism, we've created environmentally mutual zones at the boundaries of these two settings.
Order Zone
Transition Zone 1
Transition Zone 2
Disorder Zone
Aesthetic: Order Zone
Aesthetic: Disorder Zone
The required items are distributed equally among four stores located in these areas as follows:
Ordered Zone - 1 store, 3 items
Disordered Zone - 1 store, 3 items
Transition Zone 1 - 1 store, 2 items
Transition Zone 2 - 1 store, 2 items
To acquire items owned by stores, players can spend in-game currency earned through playing mini-games or choose to steal them. Players face time pressure, leading to real-time decision-making. In order to restore the constraints of rules and morals in real life, players will receive warnings if caught stealing, resulting in time penalties, thus creating a risk-reward system for theft.
.png)
Options: Buy or Steal
Mini Game: Bollon Shooting
Research Objective: We are interested in exploring whether players are more likely to steal from the disordered environment rather than the ordered one.
2.2 Experiment Variables
Our goal is to measure the effect of the level’s presentation on the player’s decision to commit the non-violent crime of theft. To gather that insight, we plan to observe the following dependent variables:
Number of items stolen or purchased by area - If the players' theft decisions differ, then we will have reason to believe they are affected by the level design.
Value of items stolen or purchased by area - If the value of items which players stolen of items differ, then we will have reason to believe they are affected by the level design.
Time spent earning money by area - If we give the player the option to make money in each area, they might develop superstitions about minigame payouts based on the level design (i.e. they may feel like they are making money quicker in the ordered area)
In order to isolate this effect, we plan to hold the variables presented in the next section constant. We plan to hold some variables constant between the two areas, to hermetically seal the appearance of the level as the variable of interest. Our independent variables are as follows:
Chance of getting caught - We established logic that would allow the player to steal twice before being arrested on the third steal. This was to teach the player that getting arrested was a mechanic in the game. After they were arrested once, the logic ensured they never were arrested again, thereby encouraging players to make subjective decisions.
Number of scavenger hunt items in each area - The player will need to collect an equal number of items in the order and disorder areas, such that they do not feel more pressure to steal in one area than another.
NPC Behavior - Shopkeepers will leave their post unattended for equal amounts of time between the ordered and disordered areas, to ensure equal opportunity to steal in both settings.
Earning Rate - The player will earn the same amount of money for playing the minigame in both the ordered and disordered areas, to simulate equal income in both areas.
2.3 Research Methods
Pre-Survey
We issued a survey collecting demographic information about our players. We are specifically interested in their frequency of playing RPG or similar video games. People who play a lot of games where non-violent crimes are available for the player to commit may be desensitized to committing them in this experiment.
Log Data
While the player is playing the 3D virtual experience, we used log data of all dependent variables listed above in an effort to measure the effect of the level’s orderliness on player behavior.
By analyzing these results, we can begin to assess the influence of our level design on the player’s willingness to commit a non-violent crime of theft. However, we will need to use qualitative methods through surveys to get a clearer picture of why the player’s behaved the way they did.
Post-Survey
We issued a post survey that asks qualitative questions about why the players made the decisions they made. For example, we have included questions regarding specific example item, including their decision (to steal versus buy), motivations, explanations, etc. In answering this series of questions, we can uncover if the player had been influenced by the level design to commit a crime they otherwise would not have committed.
3. Test Results
Total Participant Number:36 (N=36, M=20, F=16)
Nonparametric test (Mann-Whitney Wilcoxon test)
We identified the dependent variables (theft attempts, theft value, purchases, purchase value) as non-normally distributed variables based on histograms and QQ plots. Therefore, the nonparametric Mann-Whitney Wilcoxon test will be used to determine whether there is a significant difference in the means between the two treatments.
After analyzing variables such as items stolen by region and the value of stolen items, and the value of goods and items purchased by region, we did not observe the impact of differences in game environment settings on player behavior.
.png)
Left: Numbers of Items Stolen by Area
Right: Numbers of Items purchased by Area
Parametric testing (T-test)We have identified time by region as a reasonably normally distributed variable. Therefore, we used a parametric two-sample t test to determine whether there was a significant difference in means between ordered and unordered regions. When the alpha level is 0.05 and the p-value is 0.8278, we cannot reject the null hypothesis that the true difference in the means of Time Spent in Region is equal to 0. In other words, we did not observe a significant difference in the time spent in a region between ordered and disordered regions.
variable correlation
The disguised related icons did not provide strong support for our hypothesis. But we noticed that the more time subjects spent in the ordering area, the less money they spent. The more time subjects spent in the neutral zone, the less they stole.
4. Reflection
In player feedback, we also noticed the design issues of the two contrasting environments we created. The unification of color and light blurs the player's perception of the environment. A more contrasting immersive experience should be added, such as NPC activities and environments. Sound etc. In addition, we learned from the survey that most players can clearly understand that they are playing games in a virtual world, and the decisions they make will not have any impact on their real life. Thus, we find that anonymity and virtual experience result in a lack of rules and moral restraints. Research also shows that a large number of games about violence and crime may also increase players' acceptance of the behavior. In the future, we may need to select non-stealing but more close-to-life behaviors for further research.
Phase 2
5. "Theory of Mind"
Theory of Mind (ToM) refers to the ability to attribute mental states—such as beliefs, intentions, desires, and emotions—to oneself and others. It is the capacity to understand that different individuals have their own thoughts, perspectives, and feelings, which may differ from one's own. This cognitive skill allows individuals to interpret and predict the behavior of others based on their mental states, facilitating social interactions, empathy, and the ability to navigate complex social situations. Theory of Mind plays a fundamental role in human social cognition and is essential for understanding and engaging in social relationships.
6. Experiment Design
The game rules and game loop remain unchanged, and the player's task is still to collect all items through purchase or stealing. We plan to study player decisions and behavior by creating conditions that allow players to empathize. We still monitor player behavior data through qualitative surveys and log data, and analyze the results.
6.1 Game Design
We used the values conveyed by two different scenes to shape the seller NPC in the store, using a more character-based model. In the sports store in the orderly area, we shaped the seller NPC into a rich and proud character; while in the convenience store in the disorderly area, we shaped the seller NPC into a pitiful and poor character. Players will be forced to have a conversation with the seller NPC when entering the store to understand their story. Players can feel that the rich seller NPC is arrogant because of their wealth, and also learn that the poor seller NPC is struggling with the cost of his son's surgery and has not eaten for many days.
Aesthetic: Rich Shopkeeper NPC
Aesthetic: Poor Shopkeeper NPC
Research Objective: We are interested in exploring whether players are more likely to steal from the rich NPC rather than the poor one.
6.2 Experiment Variables
During this phase, only the participants’ decision-making data need to be collected, which is Number of items stolen or purchased by area/store - If the players' theft decisions differ, then we will have reason to believe they are affected by the NPC design.
6.3 Research Methods
Pre-Survey
We issued a survey collecting demographic information about our players. We are specifically interested in their frequency of playing RPG or similar video games. People who play a lot of games where non-violent crimes are available for the player to commit may be desensitized to committing them in this experiment.
Log Data
We used log data of players' decision made. By analyzing these results, we can begin to assess the influence of our NPC design on the player’s willingness to commit a non-violent crime of theft.
Post-Survey
We issued a post survey that asks qualitative questions about why the players made the decisions they made. For example, we have included questions regarding specific example item, including their decision (to steal versus buy), motivations, explanations, etc. In answering this series of questions, we can uncover if the player had been influenced by the NPC design to commit a crime they otherwise would not have committed.
7. Test Results
Total Participant Number:10 (N=10, M=6, F=4)
After analyzing variables of items stolen by region and the value of stolen items, and the value of goods and items purchased by region, we did observe the impact of differences in NPC settings on player behavior. We found that more theft occurred in neutral shops where the shopkeepers had no identifiable appearance and dialogue system. Comparing affluent and poor stores, poor stores had less theft. Comparing the overall theft rates, we found that poor-quality shops with poor-quality shopkeepers had the lowest theft rates, rich shops with rich shopkeepers had higher theft rates, and neutral shops had the highest theft rates.
.png)
.png)
Qualitative Feedbacks
Post survey data showed that all ten participants noticed the differences in stores and conversations with NPCs, and nine participants expressed a high level of empathy and expressed their willingness to use money to buy items from poor stores.
8. Reflection
Due to the small number of participants, we can only obtain preliminary guesses about the existence of theory of mind in virtual experiences, and cannot prove that this guess is completely established. At the same time, the design should be improved to better reflect the character and situation of NPCs to provide a stronger and more reliable emotional impact.