Abstract:

Experiments on real-world behaviour face a trade-off between ecological validity and experimental control: stimuli and task shall be as realistic as possible, while all relevant variables shall be carefully controlled. In my talk, I will discuss how we use different versions of virtual reality (VR) to address this challenge. The starting point will be scene-viewing experiments – how do image properties influence where people look at and how long they dwell there? Using these examples, I will argue that there are fundamental limitations to the inferences that can be drawn from typical laboratory experiments about natural gaze behaviour. Among others, these relate to the limited field of view, the absence of head and body movements and the lack of environmental constraints. To demonstrate how VR helps to overcome these issues, I will present studies that combine high-fidelity virtual versions of actual buildings with a realistic control of head and body rotation in VR. This setting allows us direct comparisons of (gaze) behaviour in the real and the virtual world, and then to perform realistic, yet well-controlled experimental manipulations in VR that would be unsafe or impossible to implement in the real world. By combining VR with walking on a treadmill, we assess how environmental constraints, such as maintaining safe gait, impact gaze behaviour, and how these effects are modulated by age. Finally, I will present some applications of this research – such as supporting the development a helmet display for firefighters – and will provide a first glimpse as to how we will extend our VR to study incidental encounters in public spaces.