Thesis

When an individual violates a norm, they infringe one or more principles of proper conduct, presenting behaviours that should not be accepted in a society. However, there are studies showing that norm violators are afforded and perceived with more power than norm abiders. To understand the When, Why and How of these findings, we implemented a video game research tool. This dissertation describes the development process of a configurable resource-management first-person multiplayer game, where players are able to follow or violate norms during resource collection and transactions. In the game, there is one leader responsible for taking or giving power to other players, within actions such as the distribution of resources and the selection of the following leader. We conducted an experiment with 20 participants to verify if the created tool was in line with prior findings. Subjects played the leader role and interacted with two confederates, a norm violator and a norm abider. We measured power perception and affordance given subjects’ game actions and answers from a questionnaire. We found results that contradicted prior studies. Only 35% of subjects selected the norm violator as the leader. Additionally, during resource distribution, subjects favoured the norm abider compared to the norm violator. Given these results, we realized that the scenario of our experiment was unbalanced - the norm violator’s scripted behaviour was extremely selfish compared to the norm abider. Even so, we noticed that a few subjects still perceived the norm violators as more skilful and, therefore, more worthy of power.

This work analyzes the feasibility of using procedural generation to create challenges in a video game based on the player's choices, such as weapon choice, and to compare that approach to one based on the player's skill as well as one based on generating content randomly. Few games have attempted to procedurally generate ways for the player to progress through the game, by generating challenges that keep the player learning new ways to use the existing mechanics. This work attempts to expand upon those concepts by three different ways of tailoring content to the player. We built a video game that generates content procedurally using the 3 aforementioned approaches and had several users test 3 different versions of the game, one for each approach. Our results suggest that, in this particular implementation, players preferred playing the random approach to the approaches with content procedurally generated, which leads us to believe that more work needs to be done to better understand how player adaptation needs to be implemented to improve play experience.

This work aims at helping developers of interactive software to test collaboration inducing scenarios. When creating a training simulation for team building, developers must make sure that their scenarios promote collaboration but also, don't force it, meaning a scenario must allow users to behave freely, otherwise did they really collaborated or were just forced to? This creates a difficulty, how can developers test their scenarios on their capability of allowing different behaviours? Our approach is based on using two different automated agents behavioural traces, one specifying the scenario's Design Goal, collaboration, and the other an example of a non-Design Goal, acting individually. After training said agents and by comparing the agents optimal behaviour when solving each scenario to the two policies, we can determine if the scenarios allow to differentiate between the Design Goal and the non-Design Goal. With this approach we are also able to order the scenarios from easiest to differentiate to hardest. Our approach was tested in two different environments, in a custom built simulator and in the iv4XR game Lab Recruits.

Due to the exponential growth of computer technologies, video games are becoming more complex each passing year; with tasks and challenges that, very often, defy the player's cognitive abilities. Handling limitations of the Working Memory and proper Cognitive Load management is crucial when dealing with problem-solving tasks; however, these concepts appear to be highly undervalued, or even unknown, in the gaming industry. To address this problem and help game designers to better understand the intrinsic complexity of their games, this work applies the attention-shifting principles of the Time-Based Resource Sharing (TBRS) Memory Model in the game Way Out (a game we have developed from scratch). We formulated the idea of Attention-Grabbing Events and tried to incorporate them into the game, aiming to create a tool-set that estimates the player's Cognitive Load while playing a video game. To validate our hypothesis, we compared the data collected from the game with the questionnaire NASA TLX -- a subjective method that assesses the mental workload experienced during a task. Although we were unable to directly estimate the player’s Cognitive Load, we believe that this work was a step forward towards achieving that goal. The amount of Attention-Grabbing Events and gameplay time, when compared with the NASA TLX, seem to be a good indicator of Cognitive Load levels. However, the TBRS Cognitive Load formula, in its current form, does not appear to be reliable when directly applied in a general gameplay scenario -- at least following the approach we did.