Wayfinding, Virtual Reality, and Cognitive Psychology
Wayfinding is everywhere. Whether planned meticulously by an urban designer or initiated informally by layers upon layers of footsteps, the spaces frequented by humans are full of information about how and why we move. To learn more about how wayfinding works, I designed and deployed an experiment in a virtual city created with video game design software.
I first started to think critically about wayfinding while working as an intern at the University of Illinois at Urbana-Champaign’s Research Park in 2019. As part of my internship, I collaborated with Research Park administration and stakeholders from across campus to brainstorm ways to make the University’s landmarks more prominent and accessible to visitors. My supervisors, who were also heavily involved in promoting tourism in the area through their cooperation with a local non-profit called Visit Champaign County, wanted to enable tourists in town to discover local destinations using a playful yet subtle wayfinding system. Capitalizing on the relatively high population density and strong transit network in Champaign-Urbana, and especially around the university campus, they wanted visitors to feel confident getting to and from local destinations without cars — using solely the bus and active transportation. Together, we imagined implementing a constellation of signs that would facilitate navigability and spark appeal throughout town.
Our vision wasn’t quite clear enough to implement before my internship ended in August of that year, but I continued to think broadly about wayfinding in the context of tourism and economic development. The following semester I began graduate school and honed my curiosities. As I developed a deeper interest in transportation planning and equity, I wanted to find ways to combine wayfinding with accessibility and mobility justice.
Illustrations of Kevin Lynch’s five elements of wayfinding, taken from Lynch’s book The Image of the City
Design
Keen on exploring further, I got in touch with Frances Ranxiao Wang, a professor in the Department of Psychology at the University of Illinois at Urbana-Champaign, who studies the ways humans solve visual and spatial problems. Professor Wang outlined the logistical steps necessary to assemble a psychology research study that would allow us to explore the mechanics of wayfinding in urban settings and ultimately make connections to accessibility and mobility justice. At this point, I had some general sketches of how I wanted the experiment space to look and feel, but I didn’t yet have the tools to bring it to life.
We knew the experiment would need to have two sections: one for testing time perception and the other for testing visual perception. We decided the first section should resemble a long urban corridor and shouldn’t allow subjects to make left or right turns. The aim of the first test section would be to investigate whether subjects’ perceptions of time were altered by the level of information provided on wayfinding signs along the virtual street corridor. We planned the second section of the test along a grid made up of several square-shaped city blocks. Within the grid, subjects would be able to make turns and navigate based on wayfinding information and a set end goal that would be communicated to them in the test instructions. The aim of the second section would be to test how effective different types of wayfinding signs are in helping subjects navigate to a landmark within the virtual environment.
Of the above images, the top two show examples of signs shown to subjects as they move through the virtual environment. The bottom two images display pop-up prompts that appear randomly throughout the duration of the experiment, regardless of “where” a subject is inside the environment. In the experimental design phase, our expectation was that by noticing patterns in sign information, like the distance to the library on the current sign versus that on the last sign or the presence of a directional arrow, subjects would be able to tailor their answers to the questions shown on the pop-up prompts.
Early sketch of the virtual environment layout
Development
Later that semester, I met an instructor of UIUC’s CS 498 VR class. Every semester, CS 498 VR students learn to use virtual reality tools to eventually develop virtual environments for selected project sponsors from departments across campus. Among other tools, students in the course learn the basics of video game design in Unity, a game development platform, while providing the University’s research community with customized virtual environments for test subjects. As one of the selected project sponsors for the Fall 2019 semester, I led a group of CS 498 VR students with the goal of producing a virtual environment in Unity that would fit the parameters of the experiment I had designed with Professor Wang.
By the end of the semester, our had group produced the virtual environment for section one of the study, whose aim was to investigate whether subjects’ perceptions of time were altered by the level of information provided on wayfinding signs along a virtual street corridor. The resulting environment included a grid of nine square-shaped city blocks and an additional segment of one street corridor with various-sized street blocks on either side. The environment also included model buildings populating each block, sidewalks along each street, and standard streetscape elements like light posts and trash cans.
Aerial view of the virtual environment early in development
Deployment
In the meantime, I recruited several undergraduate psychology students from the department’s research fair to take on the front-end design of the virtual experiment. Thanks to some excellent undergraduate research apprentices from the department, we were able to complete a round of pilot tests before Covid-19 lockdown threw our research plan into a whirlwind. In exchange for helping out, the students earned credit for an open-ended research course offered by the department. Here are the basic conclusions we drew from initial pilot testing, which we presented at UIUC’s Undergraduate Research Symposium:
Subjects tend to completely discount signs they pass in the virtual environment.
When subjects did notice a sign in the virtual environment, they tended not to translate the spatial information on the sign into temporal information.
Subjects were not able to guess the intended purpose of the experiment upon completing the test. Once the back-end programming and design work was complete, the psychology students took over and began integrating instructions, text prompts, and a debriefing page.
Research poster produced by URAP students for presentation at UIUC’s Undergraduate Research Symposium in 2021 (click to enlarge)
Then the infamous week of March 9th, 2020 hit… Since most of the in-person experiments in UIUC’s Department of Psychology are run using undergraduate psychology students as subjects, several PhD students from the department had long been conducting experiments remotely to reach larger and more diverse subject pools. They assisted Frances and I in devising a plan to migrate the existing experiment online using Pavlovia so the experiment could continue to be run remotely online during lockdown. It was too late to begin online migration that semester so I began looking for opportunities to recruit students for help in the fall 2020 semester.
Thanks to fellow grad student-in-need-of-research-disciples Dan Plyukhin, I came across UIUC’s Undergraduate Research Apprenticeship Program (URAP). The program allows graduate students actively conducting research to interview and recruit up to two undergraduate students from related disciplines to learn about and assist with the work for one semester. The application and recruitment phases of URAP lasted the entire fall semester but by spring of 2021 I was ready to commence the work of migrating our virtual experiment online with the help of two students, one from the Department of Psychology and the other from the Department of Statistics.
By taking a video screenshot of a complete run-through of the experiment in Unity, we were then able to split the video into smaller clips which could be uploaded to Pavlovia and combined with an interactive informed consent page, instructions, buttons, text prompts and a debriefing page. Where before Covid-19 lockdowns we had planned to observe participants and keep track of their timing and movement on paper, we were now able to automate that work using PsychoPy and by linking the Pavlovia experiment with a live Excel spreadsheet that kept track of results and auto-updated when each subject finished their test.
Streetscape view of the near-complete virtual environment
That work led us to another round of all-online pilot testing which is still underway. Scroll down to find out how you can participate in the project or learn more!
A walk-through of the virtual environment produced by CS 498 VR students in Unity
Finally, a huge THANK YOU to everyone who helped with this project: Laura Weisskopf Bleill (University of Illinois Research Park), Frances Ranxiao Wang, Anna Madison, Chris Widdowson, Rose Meacham, Hae-In Kim, Johnny Woo, Yinuo Peng (UIUC Department of Psychology); Daniel Cermak, Eugenia Chen, Brian Yang, Carl Guo, Michael Wan, Anna Kim, Dan Plyukhin (UIUC Department of Computer Science); Jim Wentworth (UIUC Innovation Studio); and Edith Yao (UIUC Department of Statistics).
Extras
Participate in the experiment here!
PsyToolkit is a great resource for researchers looking to build an all-online, test-based research study.
Click here to view a free PDF of Kevin Lynch’s landmark book (pun intended), The Image of the City, which served as the basis for this project.
Further reading #1: Perception and estimation of time by P. Fraisse (1984)
Further reading #2: Properties of the Internal Clock: First- and Second Order Principles of Subjective Time by M. J. Allman, S. Teki, T.D. Griffiths, and W. H. Meck (2014)
Further reading #3: Comprehending the roles of traveler perception of travel time reliability on route choice behavior by Z. R. Moghaddam, M. Jeihani, S. Peeta, and S. Banerjee (2019)
Further reading #4: Disatance Estimation of Cognitive Maps by P. W. Thorndyke (1981)
