Abstract
We studied spatial navigation (wayfinding), focusing on the decisional processes by which subjects decide what path to take through a complex environment. In addition to spatial factors as usually emphasized in the literature, we focused on “subjective” components, meaning social or functional valences that people attach to parts of the environment. To understand how people are affected by space semantics, we built a virtual hallway that our subjects could navigate and explore through a desktop computer. The hallway featured a desk that had a different appearance depending on the assigned condition: participants may see a bare desk, or a desk covered by objects in use, or a desk currently being used by an agent standing behind it. We also modified the amount of space between the desk and the adjacent wall, and the participant’s starting point (closer to the wall or far away from it). We then tested how these non-spatial and spatial cues affected people’s navigational choices - specifically how willing they were to go behind the desk (i.e. on the side where the desk user might be). We fitted a multinomial logistic Bayesian regression evaluating the factors predicting the participants’ choices. We found that purely spatial factors such as the participant’s starting point (F(2)=55.25, p<0.0001) and the table position (F(2)=21.78, p<0.0001) were weighed more heavily than the semantic qualities of the environment, including the table’s appearance (F(2)=9.89, p=0.0001). However space semantics did play a role: subjects tended to avoid area semantically marked as “in use” by another person, but only up point in the point at which geometric conditions made it extremely inconvenient to avoid. We conclude that space semantics play a role in wayfinding, albeit one secondary to ordinary spatial factors.