Using Lighting to Enhance Wayfinding

Shawn Blake James Hall Samantha Sissel

Industrial Engineering Computer Science Industrial Engineering

Abstract tell the player what is happening in the world but uses
Gaming is very popular in today’s society and, at times, clues and visual cues in the environment to allow the
developers of such virtual environments attempt to mimic player to guess what is happening and make up his own
the real world. Because lighting is important in video conclusions. This is a push towards a constructivist
games as well as in the real world, this paper attempts to method of explaining the virtual world inspired by Jean
determine if there is a significant difference between the Piaget’s writings. The idea is that by allowing the player to
decisions in which people will travel based on different determine what is happening through exploring the
wayfinding techniques. This was presented to participants environment, it will build investment in the game and draw
through a study in which they made a decision after them further into the story. This does not limit itself to the
viewing each screenshot of a virtual environment. We plot of the game but also traversing the environment. The
found that there was a significant difference between the goal is to invite players to interpret situations based on
number of people choosing to go toward the light. And we their views and experience. Developers want to use this
concluded that signage does play an important role in the idea to help players navigate the environment. They want
decision-making process; however, more people will the player to survey the environment and, using
choose to traverse a lit hallway rather than an unlit one. environmental cues, determine the best way to travel. Not
everything should be given to the player, only hints that
1 Introduction allow them to determine the best path.
Virtual Reality (VR) is being used more and more to
simulate real world experiences, as well as for training and In the case of environmental cues, the Law of Closure
gaming purposes. Recently, a gamer posted a blog says that as humans we have an innate need to
regarding the presence and absence of lighting and its categorize and fit visual elements into a larger framework.
impact as navigational cues to other gamers [Lundeen We draw conclusions based on what we can see and fill in
2009]. It is a general belief that increased levels of what we hope to be the rest. If a player sees light coming
illumination result in performance improvement. The from a door, the developer hopes that the player will come
purpose of this study is to verify this belief by comparing to the conclusion that there is a light source or way to the
which wayfinding technique is more significant, lighting or outside on the other side of the door.
signage. Wayfinding is the art of using landmarks,
signage, pathways, and environmental cues to help first- Wayfinding techniques have been used to help guide
time visitors navigate and experience a site without people through unfamiliar environments for decades.
confusion [Biesek Design 2010]. It is important to study Darken and Sibert’s research [1996a] suggests that virtual
this, not only to improve video games, but also for the world navigators may wander aimlessly when attempting
improved design of buildings and city streets at night. to find a place for the first time. This is due to problems
associated with wayfinding being used incorrectly or not at
2 Background all within the virtual environments. Therefore, using
Many applications of virtual environments (VE’s) require directional cues will minimize disorientation [Darken and
people to have some kind of spatial knowledge and use Sibert 1996b]. Wayfinding can also be described as the
visual cues to perform well in the VE [Vembar et al. 2003]. process of using visual information to find your way
Recently there has been some discussion on the internet through a complex environment [Biesek Design 2010].
blogs about how lighting is used to help guide people to
where they are supposed to go next in the virtual One of the most classic light cues experiments that have
environments [Lundeen 2009]. Visual cues have been been done was by Taylor and Socov [1974]. The study
considered as one of the most effective ways of improving was designed to show how many people would navigate
human performance [Vembar et al 2003]. around a barrier using light cues. When both sides of the
barrier had the same amount of illumination it was found
According to Smith and Worch [2010], a game design that 69% of people went to the right side of the barrier.
technique that developers are pushing for now is When the left side of the path had a higher level of
environmental storytelling. The developer does not directly illumination 75% of the people went to the left, towards the
light. The study concluded that people were attracted by
higher illumination, and therefore chose to follow the
brightest path.

Yorks and Ginthner studied the effects that wall lighting

has on desk selection [1987]. They had participants walk
into a room that had three desks inside, one near the door,
one in the middle of the room, and one at the far side of
the room. When the illumination of the room was equal,
subjects chose to sit at the desk closest to the door.
However when the back wall had a higher illumination
level than the rest, the participants chose to cross the
room and sit at the desk next to the illuminated wall. Again Figure 1: Control image of a hallway (both sides lit, no
this study shows that people are attracted to higher sign)
illumination and will walk farther distances to get to the
brightest area of the room.

Lack of further research and more consistent findings has

been one of the reasons that our understanding of the
relations between behavioral, subjective, and visual
reactions, and, which aspects of the physical environment
cue them, still remains primitive [Antonakaki 2006]. Most
of the research that has already been done dealing with
using lighting cues has not been in virtual environments.
There is also a limited amount of research done on
usability and adaptability of virtual environment interfaces.
By integrating use of an eye tracker, this experiment will
be able to present data on participants’ fixations and
Figure 2: Hallway environment with lighting on the left hall
compare it to their unique decisions.

3 Hypothesis
People will choose a lighted path when navigating a virtual
environment.

4 Method
a. Participants
There were twenty participants who completed this study.
There were seven females and thirteen males.

b. Stimuli
Stimuli in this study consist of sixteen screenshots from a
virtual environment. These screenshots are from the same Figure 3: Hallway environment with right side lit and arrow
location, an evenly divided hallway, split by a barrier. left
Screenshots differ as a result of various lighting
techniques and other visual cues, such as exit signs and at 50 Hz with a position accuracy of ~0.5˚. We used Tobii
arrows and allow participants to view the environment Studio software to run the eye tracking experiment and
from a first-person perspective. Areas of interest were gather the data.
created around the visual cues that were added to the
environment and lighting sources when applicable.
To develop the virtual environment, we used Valve
Software’s Source SDK Hammer Editor. In order to
c. Materials and Apparatus traverse the environment and gather screenshots, we
The eye tracker used for this experiment was a non- used Valve Software’s Garry’s Mod. The environment
invasive Tobii ET-1750 attached to a TFT 17” monitor with code is compiled by the Source Engine and run within the
a 1280x1024 resolution. Data on eye position is gathered Garry’s Mod environment, allowing for editing to take
place within the environment itself. Figures 1-3 show e. Procedure
example screenshots being used in the experiment. During the study, participants were greeted and briefed on
the nature of the study using a script. Then they read an
d. Experimental Design informational document approved by the Clemson
To help understand how light, whether sunlight or artificial, University Institutional Review Board. Subsequently,
impacts the direction humans tend to go, in an unfamiliar participants were be asked to take part in the calibration of
environment, we designed a study in which participants’ the eye tracker. The ultimate task was for each participant
eye fixations can be tracked to help determine whether to look at images of an environment and to click which
participants focus on light when making a decision about direction they would like to travel at his or her own pace.
which direction to travel. Upon completion, participants were thanked and
dismissed.
This was a within group study since all participants
experienced multiple conditions. The data collected from 5 Results
the study was the participants’ eye tracking data. Our The eye tracking information was gathered from
focus was to measure the speed in which the participants participants and analyzed in Minitab. To determine if there
made a decision and what direction the participant was a significant difference in the direction (toward the
decided to travel, toward the light versus in the direction of light versus away from the light) the participants chose to
the other visual cues. We also used heat mapping to help travel, the paired t-test was used. As Table 2 shows, the
us understand what the participants focus on the most. p-value is 0.000 which means that there is a significant
Another interesting statistic included gathering information difference between the number of people choosing to go
based on the participant’s gender to determine if there toward the light versus the number of people choosing to
was a significant difference in the time it took to make a go away from the light.
decision and the decision made between genders.
Table 2: Paired T-Test and CI: toward light, away from
Screenshots were developed based on a 3 x 5 model, as light
shown in Table 1. There are four versions of the test, with N Mean StDev SE Mean
each of the 15 screenshots randomized. Each participant toward light 14 15.214 2.751 0.735
away from light 14 4.143 2.316 0.619
experienced only one of the four randomly assigned tests.
Difference 14 11.07 5.03 1.34

Table 1: 3 x 5 Model of screenshots used in study 95% CI for mean difference: (8.17, 13.98)
Both Sides Lit, Left Side Lit, No Right Side Lit, No T-Test of mean difference = 0 (vs not = 0): T-Value = 8.24
No Sign Sign Sign P-Value = 0.000
Both Sides Lit, Left Side Lit, Right Side Lit,
Arrow Left Arrow Left Arrow Left To determine if there was a difference in the time it took to
Both Sides Lit, Left Side Lit, Right Side Lit, make a decision based on whether the exit sign, arrows,
Arrow Right Arrow Right Arrow Right
or no sign was present, we used the One-Way ANOVA
Both Sides Lit, Left Side Lit, Exit Right Side Lit,
Exit Left Left Exit Left test. As shown in Tables 3a and 3b, the p-value is 0.343
Both Sides Lit, Left Side Lit, Exit Right Side Lit, which shows that the information given does not present a
Exit Right Right Exit Right significant difference in total time taken to make a
decision.
The participants’ chosen direction and the eye tracking
data were analyzed using Minitab 16, a statistical software Table 3a: One-way ANOVA: Arrow, Exit, No Signage
package. The paired t-test was used to analyze the Source DF SS MS F P
direction the participants traveled and the differences in Factor 2 3.53 1.76 1.16 0.343
gender data. The one-way ANOVA was used to analyze Error 13 19.71 1.52
Total 15 23.24
the time it took to make a decision. A scatterplot was used
to determine if there was a correlation between the S = 1.231 R-Sq = 15.18% R-Sq(adj) = 2.13%
direction the participants chose to travel and the time
spent looking at each AOI.
 Table 3b: Individual Confidence Intervals

Individual 95% CIs For Mean Based on

Pooled StDev
Level N Mean StDev ---+---------+---------+---------+------
Arrow 6 3.428 0.671 (----------*----------)
Exit 6 2.762 0.969 (----------*---------)
No Signage 4 3.951 2.063 (-------------*------------)
---+---------+---------+---------+------
2.0 3.0 4.0 5.0

Pooled StDev = 1.231

In order to see if there was a correlation between the significant difference between the genders with regard to
number of people traveling toward the light and the the amount of time taken to make a decision (Table 5).
amount of time the areas of interest (AOIs) were viewed,
we used a scatterplot. For each image, we considered Table 4: Paired T-Test and CI: Males: Toward Light,
three AOIs—the area located in the direction of the light, Females: Towards Light
the area located away from the light, and the barriers N Mean StDev SE Mean
which included any kind of signage. Figure 4 shows there Males: Toward Light 14 9.571 1.697 0.453
Females: Toward Light 14 5.643 1.499 0.401
is a relationship between the time it took for participants
Difference 14 3.929 1.639 0.438
looking toward the light and the barrier and the number of
participants choosing to travel toward the light. There is a 95% CI for mean difference: (2.982, 4.875)
strong negative correlation between the amount of time T-Test of mean difference = 0 (vs not = 0): T-Value = 8.97
people looked away from the light who chose to go toward P-Value = 0.000
the light.
Table 5: Paired T-Test and CI: Males, Females
Number of People vs Time
N Mean StDev SE Mean
A OIs
Males 16 2.866 0.983 0.246
19
Number of People Choosing Toward Light

Look ed at Barrier
Look ed Towards Light
Females 16 4.169 2.087 0.522
18
Look ed A way from Light Difference 16 -1.303 1.622 0.406
17

16 95% CI for mean difference:(-2.168, -0.439)

15 T-Test of mean difference = 0 (vs not = 0): T-Value = -3.21
14
P-Value = 0.006
13

12 6 Discussion
11 The purpose of this study was to determine whether
10 people would go toward the light. We used exit signs and
0 1 2 3 4 5
Time (seconds)
arrows to determine if they would have an effect on
participants’ decisions. We found our hypothesis to be
Figure 4: Traveling toward the light vs. time spent viewing
true—more people tend to go toward the direction of the
AOI light; however, when any kind of signage is involved,
people started to question which direction to go. This
We were interested in seeing if there was a difference in accounts for the decline in people going toward the light
the percentage of males choosing to go toward the light even though the majority still went toward the light.
and the percentage of females choosing to go toward the
light. As Table 4 shows, the p-value is 0.000; therefore, It has been determined that it did not matter what type of
there is a significant difference between the males and signage was present in the image; it still took the same
females choosing to travel toward the light. amount of time for participants to make a decision. We
found this interesting because we thought that people
We were also interested in determining if there was a would focus more and take longer on the images that had
significant difference in the amount of time it took for signage. We believed that they would question whether to
males and females to make a decision. The paired t-test use the signage or to follow the lights, causing them to
resulted in a p-value of 0.006, which shows that there is a spend more time making a decision. However, now we
believe that participants went with their first instinct;
therefore, they did not take the time to question the signage/barrier most of the time and the lit hallway a little
signage or light. Instead, they chose the first direction that less often, rarely looking toward the unlit hallway. Even
felt comfortable to them. though the sign draws more attention, participants still
chose the lit path.
As shown in Figure 4, the two images that had nineteen
participants choosing to travel toward the light (images:
light left, exit left and light right, exit right) looked toward
the light and the barrier more than the non-lit side of the
barrier. This leads us to believe that they used both the lit
hallway and the exit sign to make their decision rather
than the non-lit hallway.

The image that had fifteen participants choosing to travel

toward the light (image: both sides lit, arrow right) shows
that people tend to focus more on the barrier since both
hallways are lit. Here we would have chosen the right side
(where the arrow points) as the “towards the light” side
since that is the way we would have predicted participants
would travel (based on the direction of the sign). Figure 6: Light Left

The image for which eleven participants chose to travel

toward the light (image: light left, exit right) is the only
image that participants focused more on the non-lit
hallway than the lit hallway. They also focused on the
barrier a lot longer than either side of the hallway. This
leads us to believe that the participants had a hard time
making a decision about which direction to travel since the
light and signage were indicating different directions.

We determined that there were a higher percentage of

females who chose to go toward the light than there were
for the male participants. Figure 5 shows that there is a
difference in the amount of time it took females and males Figure 7: Light Left, Exit Right
to make a decision. Although the female data is highly
variable, females took longer, on average, to make
decisions than males did during this experiment.

Histogram of Males, Females

Normal
Variable
0.4 Males
Females

Mean StDev N
2.866 0.9831 16
0.3 4.169 2.087 16
Frequency

0.2

0.1

Figure 8: Light Right, Exit Left

0.0
0 2 4 6 8
Time 7 Limitations
We were unable to use an actual virtual environment.
Figure 5: Time it takes to make a decision (by gender)
Tobii Studio required us to strictly use screenshots instead
of a continuous video stream in order to reduce the
We also used heat maps to visually determine what number of variables among participants. Using a virtual
participants viewed. The following series of heat maps environment, in which participants could actually move
(Figures 6-9) verifies that participants looked toward the
around would have been more realistic; therefore,
participants’ emotions may have played a greater factor BIESEK DESIGN. 2010. Definition of Wayfinding.
into the decision-making process. http://www.biesek.com/whatis.html

DARKEN, R. P., AND SIBERT, J. L. 1996. Wayfinding

Strategies and Behaviors in Large Virtual Worlds. In
Proceedings of the SIGCHI conference on Human
Factors in Computing Systems: Common Ground,
ACM 142-149.

DARKEN, R. P., AND SIBERT, J. L. 1996. Navigating Large

Virtual Spaces. International Journal of Human-
Computer Interaction, 8,1 (Jan./Feb./Mar. 1996), 49-
72.

LUNDEEN, R. 2009. L4D Art Direction Part 2: Stylized

Figure 9: Light Right, Arrow Left Darkness. In Left 4 Dead Blog.

8 Future Work SMITH, H. AND W ORCH, M. 2010. What Happened Here:

Future research would involve video-taping participants in Environmental Storytelling. In Proceedings of Game
action and having them retrospectively “think aloud.” This Developers Conference 2010.
would ensure that accurate time data can still be collected.
“Think aloud” could have assisted us with determining why TAYLOR, L. H., AND SOCOV, E. W. 1974. The Movement of
participants made a specific decision instead of People Toward Light. Journal of IES, (Apr. 74), 237-
speculating. A few times, we observed participants who 241.
we believe purposely chose to “rebel” and go toward the
unlit side. VEMBAR, D., IYENGAR, N., DUCHOWSKI, A., CLARK, K.,
HEWITT, J. AND PAULS, K. 2003. Effect of Visual Cues on
Since we did find a difference between genders, we think Human Performance in Navigating through a Virtual
it would be interesting for more research to be conducted Maze. In Proceedings of Eurographics Symposium on
to see if this is a common trend. Virtual Environments 2004.

And since we were limited to screenshots instead of an Yorks, P., and Ginthner, D. 1987. Wall Lighting
actual virtual environment, future work should be done to Placement: Effect on Behavior in the Work
Environment. Lighting Design and Application, 17, 7
develop a program that supports a gaming environment. It
(Jul.), 30-37.
would be interesting to see how participants react to
various obstructions in the environment and what impact
this would have on time to make a decision and which way
the participant would choose to travel.

9 Conclusion
We determined that our hypothesis, that more people will
choose the lighted path, is accepted based on our
experimental results. We found that even though signage
played a factor in the decision making process the majority
of the participants still went towards the lit hallway.
Therefore, lighting is a more significant wayfinding
technique than signage.

References
ANTONAKAKI, T. 2006. Lighting within the Social Dimension
of Space: A Case Study at the Royal Festival Hall,
London. In Proceedings of the Space Syntax and
Spatial Cognition Workshop, (Sep. 2006)