¸£ÀûÔÚÏßÃâ·Ñ

Can Video Games and Virtual Worlds Help Us Understand Architecture?

How does a space make you feel? It sounds like a question for a poet or a philosopher, but it's one that architects, psychologists, and neuroscientists have wrestled with for decades — and it's the question that sat at the heart of PhD student ’s research: Towards a Model of Affect in Architectural Experience: Using Virtual Environments as Spatial Elicitors to Capture Real-time and Continuous Observer Feedback.

The Challenge of Measuring Subjective Experiences

The idea that built environments affect our emotions is ancient. Plato wrote about it. Renaissance architects were obsessed with it. Turning into something measurable, something modelable, has proven stubbornly difficult. Because ultimately, it is a personal subjective experience that you are trying to model. Modern studies approach this by mapping spatial features — ceiling height, lighting, materiality — onto psychological axes like pleasure and arousal, following the framework psychologist James Russell proposed in the 1970s. Others go further, using fMRI scanners or EEG headsets and study how the brain responds to architectural stimuli, giving rise to the growing field of neuro-architecture.

The problem? Many of these approaches treat space as a static snapshot: nothing more than a photograph. A frozen frame. But walking through a building isn't like looking at a photo — it unfolds over time. The way a corridor narrows before opening into a grand hall, the gradual shift from harsh light to warm amber, the cumulative weight of proportion and material as you move through a sequence of rooms. These are temporal experiences, and most research simply wasn't capturing that.

Affective Computing Approach to Architectural Experience

Screenshot of the PAGAN Annotation system

Most approaches that collect psychological and non-intrusive feedback from observers do so only after they’ve experienced a room—usually as a single, static measurement. This research takes a different angle, drawing inspiration from Affective Computing, the branch of AI that explores how people emotionally respond to continuous media like films and video games. The idea is simple: why not think about architectural experience in the same way? Not as a fixed moment, but as something that unfolds, shifts, and builds meaning over time.

To test this, three different types of stimuli were used: recorded walkthrough videos, immersive virtual reality environments, and video games. Yes, video games. Interactive virtual worlds turned out to be a remarkably powerful tool — they allow participants to move freely and naturally through a designed space while researchers capture first-person emotional annotations in real time, without any invasive equipment.

Developing a Model for Architectural Experience

Across four user studies, the research built toward a core ambition: training machine learning models to predict the affective impact of architectural experience from spatial variables. Think of it as a kind of emotional equation where ceiling height, lighting colour, wall coverage, area, and spatial context combine to produce a predicted emotional response in the observer.

The implications are significant. For architects, it offers a new kind of feedback loop, one where design decisions can be stress-tested for their emotional impact before a single brick is laid. For AI and affective computing, it opens up an entirely new domain: the built environment as a dataset of human feeling. It's a rare piece of research that sits comfortably at the intersection of ancient questions and genuinely new tools. The question of how spaces make us feel is as old as architecture itself. The methods used to answer it have never looked quite like this.

VR Experiment in progress

You can read the details of Emmanouil Xylakis’s research in his Dissertation (supervised by Prof. Georgios Yannakakis and Prof. Antonios Liapis), available in the University of Malta Library.

About the Institute of Digital Games

Affective computing is one of the specialties of the Institute of Digital Games with Prof. Yannakakis’s research on the ordinal nature of emotions published in IEEE Transactions on Affective Computing proving to be an influential paper in the field. The Institute of Digital Games is ranked in the top 10 institutions active in technical games research, above technology leaders such as Google, New York University, and Georgia Tech. Moreover, they have also been recognised as belonging to the top 2% of most influential researchers in the world in Computer Science.