Overview

For my Master’s Thesis, with my group I worked on a game that compared eye tracker input with traditional input to measure player performance, immersion and engagement. The result was a kind of exploration game with an emergent narrative, where the player populates the map with objects from four distinct biomes based on what they looked at last. The hidden goal was to find 4 golden objects, one from each biome, but the players were told they could stop playing at any time before meeting this objective.

We spent a good chunk of the semester researching various projects that used eye tracking in any capacity in games or gamified projects. In most of the cases, existing games were retrofitted to allow replacement of input by eye tracking. That’s why we designed this game with eye tracking in mind but in a way that would allow for (more or less) the same experience with keyboard input.

A different take we had on the matter was using it in an emergent narrative. Most studies used game with a more typical narrative or did not focus on this aspect at all. By using an emergent narrative, we wanted to avoid the participants’ perception being swayed due to their preference in narratives and themes.

The final version used for testing embodied the vision we had when starting out well, but there is always room for improvement. If you’re interested, feel free to read the submitted paper, or the worksheets.

My tasks

Research

For this project, we only had three people in the group, so this was a good chance for me to try my hand at everything. As always, research was a huge part of this project, and since my 8th semester project also heavily focused on research, I was more than willing to undertake a big portion of this task. I looked through the relevant databases for papers that could give us an idea of the advancements and solutions in this field and could inform our game’s design. To help with this endeavour, we created a state-of-the-art table, in which we could highlight the important information from the papers to make it easier to synthesise the knowledge.

Writing

Having taken a specialised course in scientific writing, I wanted to keep exercising these muscles, so I took big part of this task as well. Writing a good paper can take a long time, with many, many, many rewrites and reiterations, and this was the case here as well. Ultimately, I’m happy with how the paper turned out, and I think this is one of my best hand-ins.

Programming

In this project, I was responsible to create the UI. Since the game was a part of an academic research, the UI did not need to be super fancy, so emphasis was on function over form. The main components were inputting participant name and number, pausing and resuming the game, and a module that captured the player’s map from above and displayed it once the player clicked on “End session” from the pause menu.

3D-modelling

All the assets in this game were created by us. We were aiming for a low-poly look, which not only allowed us to try working in this style, it also helped keep the game’s size low which became a huge plus with the remote testing. I was responsible for creating some assets, and the texturing was handled separately, as the colours of the objects were randomized for each spawn.

User testing

As with all of our projects, user testing is a crucial part. Since the Covid-19 lockdowns were in effect when user testing was carried out, we had to perform the tests remotely. This made for less than ideal circumstances. One of our conditions required an eye tracker, that very few people we reached out to had access to, so that meant we needed to organize a way to pass our own trackers around. This limited the number of participants we could attain, and the additional setup the participants in the eye tracking condition were required to do more than likely affected the time they spent completing our test. While things did not turn out perfectly, this is a good instance of having to come up with solutions to mostly unforeseen circumstances.

Data anaylsis and evaluation

I was responsible for taking a deeper look into the data the tests yielded. For this, I used rStudio and Microsoft Excel, and I have also cross-checked our results with other resources as well, to make sure the recorded values are indeed correct. This involved making sure that the correct statistical tests were employed in each step of the evaluation process.

Reflection

We went for a very minimal UI as to not have any distractions for the users, especially in the eye tracking input condition. That meant that the tracker that indicated the number of found/total golden objects were attached to the golden objects instead, visible only when these objects were in the field of view of the camera. This made it difficult to keep track of progress sometimes. An alternative solution that would have worked would have been displaying this information somewhere onscreen, but visible only when the users look directly at the area containing the information.

The biggest crux of the experiment was not telling the participants about the objective and mechanics of the game. This is really a rookie mistake that should not have happened this late into our education, but we can chalk it up to the extraordinary circumstances the testing took place in. The setup required for the eye tracker input potentially the total play time between the two conditions, as setting up the eye tracker could take up to 10 minutes, which the participants with traditional input did not have to bother with.

All-in-all, while the game might not be super exciting, but it was developed from the ground up by us, and the design and mechanics are rooted in research.