Content uploaded by Christos Mousas
Author content
All content in this area was uploaded by Christos Mousas on Jul 06, 2021
Content may be subject to copyright.
Received: 3 July 2020 Accepted: 15 July 2020
DOI: 10.1002/cav.1941
SPECIAL ISSUE PAPER
Eye fixations and electrodermal activity during low-budget
virtual reality embodiment
Claudia Krogmeier Christos Mousas
Department of Computer Graphics
Technology, Purdue University, Indiana,
Correspondence
Christos Mousas, Department of
Computer Graphics Technology, Purdue
University, West Lafayette, IN 47907, USA.
Email: cmousas@purdue.edu
Summary
As motion-sensing devices become more accessible to consumers, it is impor-
tant to understand how users experience embodiment when using these devices.
In our 3 ×2 between-groups study, we explored eye fixations and electroder-
mal activity (EDA) in order to more objectively understand potential interaction
effects between the self-avatar body, and the presence of a mirror within the con-
text of low-budget embodiment. We developed six experimental conditions con-
cerning Body (human, mannequin, and zombie self-avatars) and Mirror (mirror
and no mirror) factors, and presented participants with a virtual environment
in which they could control their self-avatars by using HTC Vive controllers and
trackers. In addition to eye fixations and EDA, we assessed self-reported data
concerning body ownership, agency, self-location, as well as enjoyment of the
experience. Our results suggest that the Body may have been more influential in
eliciting body ownership than the Mirror, and that an interaction effect between
Gender and Body may influence eye gaze behavior. Additionally, female partici-
pants reported significantly higher agency than males. We consider logical next
steps for similar research which might elaborate upon our findings.
KEYWORDS
body ownership, electrodermal activity, embodiment, eye fixations, low-budget, mirror, self-avatar,
virtual reality
1INTRODUCTION
Virtual reality research investigating embodiment, the sense of having and perceiving a virtual body as one’s own body,
often involves the use of a mirror within the virtual environment for the purposes of eliciting a sense of embodiment.1-5 In
addition to including a mirror in the virtual environment, another important factor in eliciting a sense of embodiment is
the appearance of the self-avatar: the virtual body one receives in virtual reality. While numerous studies have examined
the importance of the appearance of self-avatar bodies or body parts for embodiment,4,6-9 less research exists concerning
potential interaction effects between Body and Mirror factors (see Figure 1).
Considering that consumers may not have access to expensive software and technologies necessary for the creation
of personalized avatars, we investigate self-avatar body types which may not be similar in appearance, but are instead
readily available to consumers. In our 3 ×2 study, we presented participants with a virtual room in which they either saw
a mirror, or no mirror in front of them within the environment. All participants received one of three self-avatar body
types which included a human, mannequin, or zombie character (see Figure 2), and could control their self-avatar with
Comp Anim Virtual Worlds. 2020;e1941. wileyonlinelibrary.com/journal/cav © 2020 John Wiley & Sons, Ltd. 1of12
https://doi.org/10.1002/cav.1941
2of12 KROGMEIER MOUSAS
FIGURE 1 Top: mirror; human, mannequin, zombie. Bottom: no mirror; human, mannequin, zombie
FIGURE 2 Self-avatars
used in the experiment: human,
mannequin, and zombie
HTC Vive trackers and controllers. Participants were able to move their limbs as they liked, and experienced the virtual
environment for two minutes before filling out a short questionnaire.
In addition to the self-reported data collected from the questionnaire, participants’ eye fixations and arousal levels
were captured in order to explore sense of embodiment as it might relate to physiology. We investigated the influence
of gender within several analyses as well, and determined three primary hypotheses in our exploration of eye fixations,
electrodermal activity (EDA), and embodiment, listed below.
•RH1:Self-avatar body type will be more influential than the presence of a mirror for eliciting body ownership.
•RH2:Increased sense of embodiment will be consistent with increased arousal, as well as decreased eye gaze toward
the self-avatar.
•RH3:Females will look significantly more often and longer at their self-avatar than males.
2RELATED WORK
Placing a mirror in front of the user within the virtual environment is thought to elicit a sense of embodiment as well as
increase a user’s sense of agency.10 Studies have explored variations in the amount of time needed to display the mirror,1
as well as placement of the mirror within the user’s peripersonal space3in order to better understand ways in which
KROGMEIER MOUSAS 3of12
virtual mirrors influence embodiment. It has also been shown that the appearance of self-avatar bodies or body parts may
influence embodiment.4,6,7
While it is known that personalized, self-similar avatars can increase feelings of body ownership,11 obtaining such
self-similar avatars is not yet possible for most consumers, as the necessary equipment for the development of personalized
avatars, such as specialized software and photogrammetry scanner systems, is expensive. Considering that the low-cost
motion-sensing devices such as the HTC Vive trackers used in this study are becoming more available to consumers,
it is important to understand how different-looking self-avatars may detract or enhance feelings of embodiment within
low-budget virtual reality applications. In the current study, we consider potential interaction effects between Body and
Mirror factors, in order to understand how these factors may influence a consumer’s sense of embodiment outside of the
laboratory.
In addition to self-reported data, past virtual reality experiments have captured physiological data by using EDA,12-15
which can be indicative of high levels of arousal or stress,16 as well as eye gaze behavior such as eye fixations,17-19 which
can be indicative of not only attention,18 but anxiety,17 engagement,20 as well as social functioning.19 Virtual reality stud-
ies capturing EDA have explored public speaking anxiety,17 and posttraumatic stress disorder21 among other interests.
Concerning sense of embodiment in virtual reality, studies have previously captured EDA in response to threat to a virtual
body part,9,12 and in order to measure the influence of stress on body ownership.22 Studies capturing eye fixations have
investigated social deficits in individuals with autism spectrum disorder,13 and used eye fixations to understand reward
learning.18 However, it appears less research has captured eye fixations in order to better understand sense of embodiment.
Within virtual reality embodiment, the influence of both avatar gender and participant gender on body ownership
have been explored. Slater et al.4showed that male participants can feel body ownership of a female virtual body, while
Schulze et al.23 similarly found that participants were able to feel body ownership even if avatar gender did not match
their gender. Lopez et al.24 determined that male participants who were embodied as female avatars had significantly
higher implicit gender bias than participants embodied as male avatars. In our study, we hope to not only contribute
more objective knowledge concerning Body and Mirror factors by capturing eye fixations and EDA, but by additionally
examining potential main or interactions effects of participant gender within low-budget virtual reality applications.
3METHODS
In this section, we discuss the participants, experimental procedure, data collection as well as the virtual environment
used in the study.
3.1 Participants
As this study was part of a larger experiment investigating Body and Mirror interactions, we had previously determined
that 72 participants (12 in each experimental condition) would suffice. In reading the consent form, participants had a
general awareness of our study goals, but had no prior knowledge which Body they would be given in the virtual envi-
ronment. Our participants (41 male, 31 female) consisted of undergraduate and graduate students (49 undergraduates,
23 graduates; ages 18–32) from Purdue University, who were recruited through e-mails, posters, and word of mouth. The
participants gave informed consent, and received monetary compensation for their time. This study was approved by our
Institutional Review Board.
3.2 Experimental conditions
Six experimental groups were determined for this study, based on the factors Body and Mirror. These groups were:
human, mirror; human, no mirror; mannequin, mirror; mannequin, no mirror; zombie, mirror; and zombie, no mirror.
We employed a 3 ×2 between-groups design in order to explore interaction effects between Body and Mirror (Figure 1).
3.3 Virtual environment implementation
The virtual room used in the experiment was created with the Unity3D game engine, and is shown in Figure 3. The room
was relatively plain, so as not to distract attention away from the self-avatar body, only including a few pieces of artwork,
4of12 KROGMEIER MOUSAS
FIGURE 3 Three different perspectives of the virtual environment with the mirror
and plants around the edges of the room. We did not collect eye fixations from these background objects, instead including
them in an effort to create a more believable environment, which would likely not be empty of any surrounding objects.
All assets were downloaded from the Unity asset store. The mirror was placed directly in front of participants, so that the
full self-avatar body could be seen. For participants in the no mirror dimension, a generic painting of the same size stood
in place of the mirror. The characters used for self-avatars were downloaded from Adobe Mixamo and Adobe Fuse. We
used FinalIK with HTC Vive trackers and controllers in order to allow participants the ability to control their limbs.
3.4 Measurements
We collected participants’ eye fixations and EDA data to more objectively understand how participants experienced the
virtual environment. Our questionnaire provided self-reported data concerning sense of embodiment, and enjoyment.
Our data collection is discussed in more detail below.
3.4.1 Questionnaire
Sense of embodiment encompasses many concepts,25 three of which are sense of self-location: the perceived location of
the virtual body relative to the physical self;26 agency: the perceived ability to control the virtual body’s actions;27 and
body ownership: the sense of owning the virtual body as one owns their physical body.28 In our 7-point Likert scale
questionnaire, we included two items for body ownership, four items for agency, and two items for self-location.29 Lastly,
we included three items to measure enjoyment of the experience.30 We additionally provided a space for participants
to express their comments at the end of the questionnaire. Each questionnaire variable was averaged for our statistical
analyses, with reverse-coded questions being reverse scored. Our questionnaire was administered on screen immediately
after the virtual experience had finished.
3.4.2 Eye fixations
Using the Cognitive3D spatial analytics platform, we captured three measurements concerning eye fixation data. Eye
fixations data included fixation count (the number of different instances of fixations on a specified object), fixation length
(the total duration time for each fixation object), and time to fixation (the time at which a fixation first appeared on an
object). We assigned the self-avatar bodies and the mirror as objects from which we would like to collect fixation data.
For the no mirror dimension, we collected fixation data from the painting which stood in place of the mirror.
3.4.3 Electrodermal activity
In order to investigate participant’s arousal level, EDA peak quantity was captured during the experience. Our EDA peak
count is a sum of all arousal responses throughout the 2-min virtual experience. Because participants moved their limbs
KROGMEIER MOUSAS 5of12
FIGURE 4 A participant given the human self-avatar,
in the mirror dimension
in order to explore their self-avatar, we understand that the EDA signal may have been significantly altered by move-
ment artifacts. However, because all participants were moving in order to examine their self-avatars, we decided to do
an exploratory assessment of EDA across groups, under the assumption that each group was equally prone to motion
artifacts.
3.5 Procedure
Participants first completed a demographics survey and signed a consent form upon arriving at the lab. Next, a Shimmer3
GSR+sensor was attached to their non-dominant hand. Two HTC Vive trackers were placed on the participant’s feet, and
one HTC Vive tracker was positioned at the center of the back. Participants were given the HTC Vive controllers to use
for hand tracking, and wore the head-mounted display (HMD) for head tracking. Eye tracking calibration was completed
using the HTC Vive Pro Eye HMD before the virtual experience was started. The researcher informed participants that
they could move the self-avatar body how they would like to do so within the virtual environment. Participant were made
aware that there were no tasks within the virtual environment, and that they only needed to experience the environment,
which would last two minutes.
Participants were instructed to stand in the center of the room, without walking within the virtual space (see Figure 4).
In this way, all participants experienced the virtual environment within the same position in the room. Once participants
indicated that they were ready to begin, the researcher began the EDA recording and started the virtual reality application,
which automatically began eye fixation data collection. After 2 min, the researcher ended the virtual reality application
and terminated the EDA recording. The participant removed the HMD and completed the questionnaire on Qualtrics, an
online survey tool.
4RESULTS
We used two-way Analysis of Variance (ANOVA) to explore main and interaction effects between Body and Mirror factors,
and used three-way ANOVAs to explore interaction effects between Body,Mirror,andGender. Post hoc comparisons
were conducted with Bonferroni corrections. Normality tests were evaluated graphically using Q-Q plots of the residuals.
Pearson bivariate correlations were used to assess correlations between our eye fixation data and our self-reported data,
as well as to explore correlations between EDA and self-reported data.
4.1 Eye fixations and EDA
In this section, we report eye fixation data as well as EDA data. Our eye fixation data includes fixation count, fixation
length, and time to fixation, while our EDA data includes the total EDA responses during the virtual experience, here
referred to as EDA peak count.
6of12 KROGMEIER MOUSAS
4.1.1 Self-fixations
We used a two-way ANOVA to explore interaction effects of Body and Mirror factors on self-fixation count.Whilewe
found no interaction between Body and Mirror (F(2,66) =.438, p=.647), or main effect of Body (F(2,66) =1.085, p=.344),
we found a significant main effect of Mirror on self fixation count (F(1,66) =7.932, p<.01). Post hoc comparisons showed
that participants who experienced the mirror dimension looked significantly less times at their self-avatars (M=106.78,
SD =25.05) than participants who did not experience the mirror (M=206.56, SD =25.05). Consistent with these results,
we also found a significant main effect of Mirror on self-fixation length (F(1,66) =6.584, p<.05). Post hoc comparisons
showed that participants who experienced the mirror (M=11.91, SD =2.80) looked significantly less at their self-avatars
than participants who did not experience the mirror (M=22.06, SD =2.80). We determined no main effect of Body
(F(2,66) =.967, p=.386), or interaction effect between Body and Mirror (F(2,66) =.392, p=.678) on self fixation length.
We also investigated time to self-fixation and found a significant main effect of Body (F(2,66) =4.214, p<.05). Post hoc
comparisons revealed that participants who were given a zombie self-avatar (M=8.18, SD =4.47) looked significantly
sooner at the self-avatar than participants who were given a mannequin self-avatar (M=26.50, SD =4.47). We found no
main effect of Mirror (F(1,66) =2.443, p=.123), or interaction between Body and Mirror (F(2,66) =.804, p=.452) on time
to self fixation.
4.1.2 Object fixations
We wanted to confirm that the mirror, in the mirror dimension, was looked at significantly more so than the paint-
ing, which would indicate that participants in the mirror dimension did in fact look at the mirror, rather than simply
look forward. With our Body ×Mirror two-way ANOVA, we assessed differences in object fixation length, object fixation
count and time to object fixation. In the mirror dimension, the object refers to the mirror in the virtual environment,
and in the no mirror dimension, the object refers to the painting. We determined no two-way interaction between
Body and Object (F(2,66) =.495, p=.612), or main effect of Body (F(2,66) =1.058, p=.353), but found a main effect of
Object (F(1,66) =75.635, p<.001) on object fixation length. Post hoc comparisons showed that the mirror (M=49.52,
SD =2.85) was looked at significantly longer than the painting (M=14.48, SD =2.85).
We found no two-way interaction (F(2,66) =.084, p=.919), or main effect of Body (F(2,66) =.899, p=.412) on object
fixation count. We determined a main effect of Object on object fixation count (F(1,66) =17.566, p<.001), with post
hoc comparisons revealing that the mirror (M=156.53, SD =10.18) was looked at significantly more times than the
painting (M=96.17, SD =10.18). Lastly, we investigated time to object fixation, and determined no two-way interac-
tion (F(2,66) =.757, p=.473), or main effect of Body (F(2,66) =.091, p=.913), but discovered a significant main effect of
Object (F(1,66) =11.213, p<.001) on time to object fixation. Post hoc comparisons revealed that participants who experi-
enced the mirror (M=1.78, SD =1.10) looked significantly sooner at the mirror, than participants looked at the painting
(M=7.00, SD =1.10).
4.1.3 Electrodermal activity
While we found no two-way interaction using our Body ×Mirror two-way ANOVA with EDA peak count (F(2,66) =.450,
p=.640), we determined a significant main effect of Body on EDA peak count (F(2,66) =3.655, p<.05). Post hoc compar-
isons showed that participants with the zombie self-avatar (M=14.13, SD =1.21) had significantly more EDA peaks than
participants with the mannequin self-avatar (M=9.50, SD =1.21).
4.2 Self-reported data
We evaluated body ownership, agency and self-location aspects of embodiment as well as enjoyment of the experience in
our self-reported data.
4.2.1 Body ownership
We neither found interaction between Body and Mirror concerning body ownership (F(2,66) =.176, p=.839), nor a
main effect of Body on body ownership (F(2,66) =1.862, p=.163). Interestingly, we also found no main effect of Mirror
(F(1,66) =2.477, p=.120) on body ownership.
KROGMEIER MOUSAS 7of12
4.2.2 Agency
No interaction effect between Body and Mirror (F(2,66) =.625, p=.539), main effect of Body (F(2,66) =.818, p=.446), or
main effect of Mirror (F(1,66) =.647,p=.424) were found concerning agency.
4.2.3 Self-location
Exploring self-location, we found no interaction between Body and Mirror (F(2,66) =.048, p=.953), main effect of Body
(F(2,66) =2.896,p=.062), or main effect of Mirror (F(1,66) =.185, p=.668).
4.2.4 Enjoyment
We also investigated enjoyment, finding no significant interaction between Body and Mirror (F(2,66) =.435, p=.649),
main effect of Body (F(2,66) =.098, p=.906), or main effect of Mirror (F(1,66) =.012, p=.914).
4.3 Gender differences
We used a three-way ANOVA in order to explore the influence of Gender on eye fixations, EDA peak count, and
sense of embodiment. While we did not find any interaction effect between Body,Mirror,andGender on agency
(F(2,60) =1.022, p=.366), we determined a main effect of Gender on agency (F(1,60) =4.607, p<.05), with females
(M=5.36, SD =0.15) reporting significantly higher agency than males (M=4.94, SD =0.13) as shown with post hoc
comparisons. We found no significant three-way interaction between Body,Mirror and Gender (F(2,60) =.524, p=.595)
concerning body ownership, nor main effect of Gender (F(1,60) =3.595, p=.063) on body ownership. We also found no
three-way interaction (F(2,60) =.531, p=.591), or main effect of Gender (F(1,60) =.267, p=.607) on self-location.While
we found no three-way interaction effect (F(2,60) =1.482, p=.235) on self-fixation length, we discovered a two-way
interaction effect between Body and Gender (F(2,60) =4.995, p<.01) on self-fixation length (see Figure 5a). Post hoc com-
parisons showed that male participants given zombie self-avatars (M=29.31, SD =4.46) looked significantly longer at
their self-avatars than male participants given mannequin self-avatars (M=10.57, SD =4.18). We determined no main
effect of Gender (F(1,60) =3.439, p=.069) on self fixation length.
Looking at self-fixation count, we similarly found no three-way interaction effect (F(2,60) =1.327, p=.273), but
again found a two-way interaction effect between Body and Gender (F(2,60) =3.371, p<.05) concerning self-fixation count
(see Figure 5b), with post hoc comparisons revealing that male participants with the zombie self-avatar (M=259.81,
SD =40.81) looked significantly more times at their self-avatar than male participants given the mannequin self-avatar
FIGURE 5 Gender ×Body interaction effect on (a) self-fixation length: males with zombie self-avatars had significantly higher self
fixation length than males with mannequin self-avatars, and (b) self-fixation count: males with zombie self-avatars had significantly higher
self fixation count than males with mannequin self-avatars. H =human, M =Mannequin, Z =Zombie. Vertical axis: (a) 0–40 seconds for
self-fixation length, (b) 0–400 for self-fixation count. Blue lines denote males and red lines denote females
8of12 KROGMEIER MOUSAS
(M=110.50, SD =38.24), but no main effect of Gender (F(1,60) =3.355, p=.072) on self-fixation count. We deter-
mined no three-way interaction effect (F(2,60) =1.710, p=.189), two-way interaction effect between Body and Gender
(F(2,60) =1.619, p=.207), two-way interaction effect between Mirror and Gender (F(1,60) =.071, p=.790) or main effect
of Gender (F(1,60) =.170, p=.682) on time to self-fixation.
Lastly, we explored the potential influence of Gender on EDA peak count and determined no significant interac-
tion effect between Body,Mirror, and Gender (F(2,60) =.524,p=.595), nor interaction between both Body and Gender
(F(2,60) =.648, p=.527), and Mirror and Gender (F(1,60) =.483,p=.490). We determined no main effect of Gender
(F(1,60) =3.595,p=.063) on EDA peak count.
4.4 Correlations
We assessed correlations between our eye fixations, EDA and self-reported data, and determined one weak negative corre-
lation between self-fixation count and self-location (r=−.233, n=72, p<.049), as well as a weak negative correlation
between time to self-fixation and body ownership (r=−.263, n=72, p=.025).
5DISCUSSION
In our 3 ×2 between-groups study, we explored the presence of a Mirror and the influence of the self-avatar Body type on
low-budget virtual reality embodiment. We collected eye fixations and EDA data in addition to self-reported data from the
questionnaire. In this section, we mention how our physiological data may relate to our self-reported data, and discuss
partial evidence or lack thereof for our RH1,RH2,andRH3.
Our RH1 stated that the self-avatar body type would be more influential than the presence of a mirror for eliciting
body ownership. However, our results revealed that neither the self-avatar body type nor the presence of a mirror in the
virtual environment influenced body ownership. It is possible that because the mirror was outside of the participant’s
peripersonal space, it was too far away to make any impact on body ownership.3Our results are similar to those of Lugrin
et al.31 who reported no significant differences in self-reported body ownership between all three avatars when examining
a human, robot, and block-man avatars. Jo et al.32 also investigated three self-avatar body types for inducing body owner-
ship, finding that the cartoon-like semblance of the actual participant elicited the most body ownership, when compared
to a realistic self-avatar and a cartoon-like character in a different outfit. Considering their results, it is possible that none
of our self-avatars elicited body ownership, as none bore any semblance to any participants. However, previous works5,6
have shown that it is possible to elicit body ownership when self-avatars differ from participants. It is possible that by
including a Customization factor, we may better understand the role of customization versus similarity in body owner-
ship within future work. Additionally, our study was limited in that we only collected eye fixations from the mirror as an
entire object, and therefore do not know if participants looked more at their self-avatar reflection in the mirror, or at the
reflections of environmental objects.
It is possible that issues in our body ownership items on the questionnaire may have obscured potential differences
reported in body ownership. Unfortunately, issues in the phrasing of our body ownership questions forced us to throw out
two items, which left us with only two questions for body ownership. Gonzalez-Franco et al.29 suggest both reverse-scored
and normally scored questions for assessing aspects of embodiment, and therefore it is possible that our two final questions
were not effective in determining participant feelings of body ownership. Considering these findings, our results do not
support our RH1.
Although we could not determine an influence of Body on body ownership in analyzing our self-reported data, we
found significantly higher arousal for participants given the zombie self-avatar compared to participants given the man-
nequin self-avatar, as seen through our EDA data. Perhaps the increased arousal found for participants given zombie
self-avatars was indicative of stress or anxiety, considering that participants may prefer interacting with avatars that are
attractive.33 Therefore, participants may have been anxious or stressed by controlling the zombie self-avatar, as it was
unattractive with its ripped clothing and bloodied skin. Lin et al.8found that zombie hands were perceived as more real-
istic than robot-like hands, and so perhaps arousal due to unattractiveness may have even been amplified by increased
realism of the zombie self-avatar. Tieri et al.34 found that physical continuity of virtual limbs was important for inducing
an electrodermal response. Although they explored body ownership within the context of perceived threat to a virtual
KROGMEIER MOUSAS 9of12
limb, their results showed that a virtual hand which was physically continuous with the wrist increased body owner-
ship as opposed to virtual hands which were disconnected from the wrist. Considering that our mannequin self-avatar’s
hands were connected to the wrist via a ball-like joint, unlike a real hand-wrist connection, it is possible that EDA was
significantly lower for mannequins for this reason.
We also determined that participants with zombie self-avatars looked at their self-avatars significantly sooner than par-
ticipants with mannequin self-avatars, suggesting a relationship between time to self fixation and arousal, however these
two variables were not correlated. Our results further suggest that the sooner a participant looked at their self-avatar dur-
ing the virtual experience, the more body ownership they reported, which suggests the potential importance for timing of
self-avatar fixations in eliciting body ownership. If we consider both that looking sooner at one’s self-avatar is correlated
with increased body ownership, and that participants given zombie self-avatars looked sooner at their bodies than par-
ticipant given mannequin self-avatars, we could consider perhaps zombie self-avatars felt higher body ownership, which
would partially support our RH1. In order to better understand if the Body factor may influence self-reported body own-
ership, we suggest future studies more deeply examine the influence of distance from participant to the mirror, as well
as continue to explore relationships between self-avatar realism and arousal, and how this relationship may impact body
ownership.
Considering self fixation count, we found that the more times participants looked at their self-avatar, the less
self-location with the self-avatar body they reported. Perhaps more eye fixations toward the self-avatar decreases the
self-location aspect of embodiment, as low-budget embodiment systems may introduce motion artifacts and cause the
participant to feel taken out of the experience, and therefore less embodied. Similarly, perhaps as participants look sooner
at the self-avatar body, they may report higher body ownership because they have had more time between first viewing the
body and the end of the virtual experience, and therefore may have had time to focus on other aspects of the experience
besides issues with motion artifacts.
We were able to confirm issues with motion artifacts in our embodiment set-up based on participant commentary.
For example, one participant mentioned “The movements picked up seemed to be slightly delayed and were less than my
movements,εsuggesting problems in motion-sensing. Another participant said, “I felt as if the virtual body was a second skin
on top of my own. But I didn’t think it was my body more like a huge costume I was wearing,εdemonstrating another reason
for the lack of body ownership reported. Perhaps due to low-budget embodiment, we can conclude that as self-location
increased, self-avatar fixations decreased, which partially supports our RH2. Based on these findings, we suggest future
studies consider having participants look at their self-avatars for a predetermined length of time, and/or at a specific time
during the experience, in order to better understand the role that eye fixation count, duration, and timing may play in
determining embodiment, and if a specific eye fixation behavior might also be best suited for embodiment.
For our RH3, we explored the potential influence of Gender, and determined a significant main effect of Gender on
agency, with females reporting significantly higher agency than males. We did not expect to find differences in agency,
as our virtual environment involved no defined tasks or goals required of participants. It is possible that an increased
exploration of the self-avatar’s capabilities led to a decreased sense of agency, as more self-avatar movement would reveal
more motion artifacts. During the experiment, males appeared to experiment with different positions and rotations of the
self-avatar more so than females. In future studies, it would be important to capture movement data in order to determine
if agency was influenced by participant movement. Considering the proteus effect, which suggests that individuals may
behave as their self-representation would be expected to behave,35 perhaps females felt that their behaviors matched their
self-avatars bodies better than males, and therefore reported higher agency.
Our results further suggest that male participants given zombie self-avatars looked significantly longer and signif-
icantly more times at their self-avatars than male participants given mannequin self-avatars. This finding additionally
emphasizes differences between zombie and mannequin self-avatars, as well as an influence of gender. However, our
results cannot fully determine whether or not these differences may be due to participant or self-avatar gender, as we
matched our participants self-reported gender with the self-avatar gender. Here, we might suggest that future studies
replicate our study in order to determine if participant gender or avatar gender was more impactful. Schrammel et al.36
found that, regardless of participant gender, fixations were longer on male virtual characters than female virtual charac-
ters, perhaps due to differences in social status or expected interactions between genders. Although their study explored
interactions with virtual characters rather than viewing a self-avatar, if we were to consider that low body ownership
may have contributed to perceiving the self-avatar as one might perceive an unrelated virtual character, our results are
in line with those of Schrammel et al.36 in that male zombie self-avatars were looked at significantly longer than male
mannequin self-avatars.
10 of 12 KROGMEIER MOUSAS
An additional reason for the difference in fixations between the zombie and mannequin self-avatar may be due to
an incongruency of actions. Marschner et al.37 found that incongruent signals may increase attentional demands of the
observer, therefore, because the zombie may have been significantly more likely to be involved in aggressive actions as
often seen in zombie-filled entertainment, perhaps more attention was paid to the zombie as it made less sense with par-
ticipant expectations for movement, with no participants implementing zombie-specific movements themselves. Future
research could better inform our results by providing congruent and incongruent tasks for participants, which might
help to determine if the differences found in agency were related to self-avatar task-related congruency. It may have also
been important to collect specific game experience data from participants prior to the study, as it is possible that partici-
pants with significant experience with zombies in entertainment may have reacted differently than those not as familiar
with zombies. Lastly, if zombie self-avatars were perceived as threatening, our results would again be similar to those of
Schrammel et al.,36 who found longer fixations in response to virtual character expressions that may have been perceived
as threatening. Although participant gender did not influence self-fixations, providing no support for our RH3,ourresults
indicate interesting interactions between Body and Gender, which we hope to continue to explore in future studies.
6CONCLUSION
In this study, we collected both physiological data and self-reported data in order to understand potential interactions
between Body and Mirror within a low-budget virtual reality embodiment application. We developed six experimental
conditions concerning Body (human, mannequin, and zombie) and Mirror (mirror and no mirror) factors, and pre-
sented participants with a virtual environment in which they could control their self-avatar with HTC Vive controllers
and trackers. We discovered differences in arousal as well as eye fixations between zombie and mannequin self-avatars,
however determined no differences in self-reported body ownership. We determined that as self-fixations increased, less
self-location with the self-avatar was reported, suggesting a potential relationship between eye fixations and aspects of
embodiment using low-budget motion sensing devices. Lastly, we determined that females reported significantly higher
agency than males, suggesting that a more complex relationship between gender and agency in low-budget embodiment
applications may exist.
In summary, next steps for future research include a deeper investigation of the influence of both mirror-participant
proximity and self-avatar realism on body ownership, and experiments designed to disentangle the uses and effects of
self fixation count, duration, and timing. Additionally, studies that pursue congruent and incongruent tasks may better
inform relationships between agency and gender. Our study provided objective data from EDA and eye fixations to better
understand how consumers experience virtual reality embodiment using the low-budget motion-sensing equipment and
different-looking self-avatars available to them. It is our hope that our findings and suggestions for similar future research
may allow for the enhanced development of low-budget virtual reality embodiment applications for consumers.
ORCID
Christos Mousas https://orcid.org/0000-0003-0955-7959
REFERENCES
1. Ito R, Ogawa N, Narumi T, Hirose M. Do we have to look at the mirror all the time? effect of partial visuomotor feedback on body ownership
of a virtual human tail. Proceedings of the ACM Symposium on Applied Perception, Barcelona, Spain; 2019. p. 1–9.
2. Latoschik ME, Lugrin JL, and Roth D. FakeMi: A fake mirror system for avatar embodiment studies. Proceedings of the ACM Conference
on Virtual Reality Software and Technology, Munich, Germany; 2016. p. 73–76.
3. Preston C, Kuper-Smith BJ, Ehrsson HH. Owning the body in the mirror: The effect of visual perspective and mirror view on the full-body
illusion. Sci Reports. 2015;5(1):1–10.
4. Slater M, Spanlang B, Sanchez-Vives MV, Blanke O. First person experience of body transfer in virtual reality. PloS One. 2010;5(5):1–9.
5. Banakou D, Groten R, Slater M. Illusory ownership of a virtual child body causes overestimation of object sizes and implicit attitude
changes. Proc Nat Acad Sci. 2013;110(31):12846–12851.
6. Lin Q, Rieser JJ, Bodenheimer B. Stepping off a ledge in an HMD-based immersive virtual environment. Proceedings of the ACM
Symposium on Applied Perception, Dublin, Ireland; 2013. p. 107–110.
7. Tajadura-Jiménez A, Banakou D, Bianchi-Berthouze N, Slater M. Embodiment in a child-like talking virtual body influences object size
perception, self-identification, and subsequent real speaking. Sci Reports. 2017;7(1):1–12.
8. Lin L, Jörg S. Need a hand? how appearance affects the virtual hand illusion. Proceedings of the ACM Symposium on Applied Perception,
Anaheim, USA; 2016. p. 69–76.
KROGMEIER MOUSAS 11 of 12
9. Hägni K, Eng K, Hepp-Reymond MC, et al. Observing virtual arms that you imagine are yours increases the galvanic skin response to an
unexpected threat. PloS One. 2008;3(8):1–6.
10. Gonzalez-Franco M, Perez-Marcos D, Spanlang B, Slater M. The contribution of real-time mirror reflections of motor actions on virtual
body ownership in an immersive virtual environment. Proceedings of the IEEE Virtual Reality Conference. IEEE, Waltham, MA, USA;
2010. p. 111–114.
11. Waltemate T, Gall D, Roth D, Botsch M, Latoschik ME. The impact of avatar personalization and immersion on virtual body ownership,
presence, and emotional response. IEEE Trans Visualiz Comput Graph. 2018;24(4):1643–1652.
12. Yuan Y, Steed A. Is the rubber hand illusion induced by immersive virtual reality? Proceedings of the IEEE Virtual Reality Conference,
Waltham, MA, USA; 2010. p. 95–102.
13. Bekele E, Zheng Z, Swanson A, Crittendon J, Warren Z, Sarkar N. Understanding how adolescents with autism respond to facial
expressions in virtual reality environments. IEEE Trans Visualiz Comput Graph. 2013;19(4):711–720.
14. Krogmeier C, Mousas C, Whittinghill D. Human–virtual character interaction: Toward understanding the influence of haptic feedback.
Comput Animat Virtual Worlds. 2019;30(3-4):e1883.
15. Mousas C, Kao D, Koilias A, Rekabdar B. Real and virtual environment mismatching induces arousal and alters movement behavior. IEEE
virtual reality and 3D user interfaces, 2020; p. 626–635.
16. Boucsein W. Electrodermal activity. Berlin, Germany: Springer Science & Business Media, 2012.
17. Kim H, Shin JE, Hong YJ, et al. Aversive eye gaze during a speech in virtual environment in patients with social anxiety disorder. Austr
NZ J Psychiatry. 2018;52(3):279–285.
18. Wilson G, Bryan J, Cranston K, Kitzes J, Nederbragt L, Teal TK. Good enough practices in scientific computing. PLoS Comput Biol.
2017;13(6):1–20.
19. Choi SH, Ku J, Han K, et al. Deficits in eye gaze during negative social interactions in patients with schizophrenia. J Nervous Mental Dis.
2010;198(11):829–835.
20. Lahiri U, Bekele E, Dohrmann E, Warren Z, Sarkar N. A physiologically informed virtual reality based social communication system for
individuals with autism. J Autism Develop Disorders. 2015;45(4):919–931.
21. Norrholm SD, Jovanovic T, Gerardi M, et al. Baseline psychophysiological and cortisol reactivity as a predictor of PTSD treatment outcome
in virtual reality exposure therapy. Behav Res Therapy. 2016;82:28–37.
22. Bergström I, Kilteni K, Slater M. First-person perspective virtual body posture influences stress: a virtual reality body ownership study.
PloS One. 2016;11(2):1–21.
23. Schulze S, Pence T, Irvine N, Guinn C. The effects of embodiment in virtual reality on implicit gender bias. Proceedings of the International
Conference on Human-Computer Interaction, Orlando, Florida, USA; 2019. p. 361–374.
24. Lopez S, Yang Y, Beltran K, et al. Investigating implicit gender bias and embodiment of white males in virtual reality with full body
visuomotor synchrony. Proceedings of the ACM CHI Conference on Human Factors in Computing Systems, Glasgow, United Kingdom;
2019. p. 1–12.
25. Kilteni K, Groten R, Slater M. The sense of embodiment in virtual reality. Presence Teleoperat Virtual Environ. 2012;21(4):373–387.
26. Arzy S, Thut G, Mohr C, Michel CM, Blanke O. Neural basis of embodiment: Distinct contributions of temporoparietal junction and
extrastriate body area. J Neurosci. 2006;26(31):8074–8081.
27. Newport R, Pearce R, Preston C. Fake hands in action: Embodiment and control of supernumerary limbs. Exper Brain Res.
2010;204(3):385–395.
28. Lopez C, Halje P, Blanke O. Body ownership and embodiment: Vestibular and multisensory mechanisms. Neurophysiol Clin.
2008;38(3):149–161.
29. Gonzalez-Franco M, Peck TC. Avatar embodiment. towards a standardized questionnaire. Front Robot AI. 2018;5:74.
30. Stevens M, Moget P, De Greee MHG, Lemmink KAPM, Rispens P. The Groningen Enjoyment Questionnaire: A measure of enjoyment in
leisure-time physical activity. Percept Motor Skills. 2000;90(2):601–604.
31. Lugrin JL, Latt J, Latoschik ME. Avatar anthropomorphism and illusion of body ownership in VR. IEEE virtual reality, 2015; p. 229–230.
32. Jo D, Kim K, Welch GF, et al. The impact of avatar-owner visual similarity on body ownership in immersive virtual reality. Proceedings
of the ACM Symposium on Virtual Reality Software and Technology, Gothenburg, Sweden; 2017. p. 1–2.
33. Kim HK, Kim SH. Understanding emotional bond between the creator and the avatar: Change in behavioral intentions to engage in
alcohol-related traffic risk behaviors. Comput Human Behav. 2016;62:186–200.
34. Tieri G, Tidoni E, Pavone EF, Aglioti SM. Body visual discontinuity affects feeling of ownership and skin conductance responses. Sci
Reports. 2015;5:17139.
35. Yee N, Bailenson JN. Walk a mile in digital shoes: The impact of embodied perspective-taking on the reduction of negative stereotyping
in immersive virtual environments. Proc Presence. 2006;24:26.
36. Schrammel F, Pannasch S, Graupner ST, Mojzisch A, Velichkovsky BM. Virtual friend or threat? The effects of facial expression and gaze
interaction on psychophysiological responses and emotional experience. Psychophysiology. 2009;46(5):922–931.
37. Marschner L, Pannasch S, Schulz J, Graupner ST. Social communication with virtual agents: The effects of body and gaze direction on
attention and emotional responding in human observers. Int J Psychophysiol. 2015;97(2):85–92.
12 of 12 KROGMEIER MOUSAS
AUTHOR BIOGRAPHIES
Claudia Krogmeier is a Ph.D. student at Purdue University in West Lafayette, Indiana. Her
research interests include virtual reality, human–computer interaction, physiology in entertain-
ment, and personalized storytelling within videos and virtual reality. Claudia holds a master’s
degree in computer graphics technology from Purdue University and a bachelor of arts degree
from Indiana University, where she practiced video production. She is open to both research and
creative collaborations.
Christos Mousas is an Assistant Professor at Purdue University, West Lafayette. His research
revolves around virtual reality, computer graphics, computer animation, and human–computer
interaction. In the past, he has been affiliated with the Department of Computer Science at Dart-
mouth College as Postdoctoral Researcher, and with the Department of Cultural Technology and
Communication of the University of the Aegean as Research Assistant. He holds a Ph.D. in Infor-
matics and an M.Sc. in Multimedia Applications and Virtual Environments both from the School
of Engineering and Informatics of the University of Sussex. He is a member of IEEE.
How to cite this article: Krogmeier C, Mousas C. Eye fixations and electrodermal activity during low-budget
virtual reality embodiment. Comput Anim Virtual Worlds. 2020;e1941. https://doi.org/10.1002/cav.1941
