Available via license: CC BY 4.0
Content may be subject to copyright.
Effects of Acrophobic Fear and Trait
Anxiety on Human Behavior in a Virtual
Elevated Plus-Maze
Octavia Madeira
1
*, Daniel Gromer
1
, Marc Erich Latoschik
2
and Paul Pauli
1
,
3
1
Department of Psychology I, Biological Psychology, Clinical Psychology and Psychotherapy, University of Würzburg, Würzburg,
Germany,
2
Department of Human-Computer-Interaction, University of Würzburg, Würzburg, Germany,
3
Center of Mental Health,
University of Würzburg, Würzburg, Germany
The Elevated Plus-Maze (EPM) is a well-established apparatus to measure anxiety in
rodents, i.e., animals exhibiting an increased relative time spent in the closed vs. the open
arms are considered anxious. To examine whether such anxiety-modulated behaviors are
conserved in humans, we re-translated this paradigm to a human setting using virtual
reality in a Cave Automatic Virtual Environment (CAVE) system. In two studies, we
examined whether the EPM exploration behavior of humans is modulated by their trait
anxiety and also assessed the individuals’levels of acrophobia (fear of height),
claustrophobia (fear of confined spaces), sensation seeking, and the reported anxiety
when on the maze. First, we constructed an exact virtual copy of the animal EPM adjusted
to human proportions. In analogy to animal EPM studies, participants (N30) freely
explored the EPM for 5 min. In the second study (N61), we redesigned the EPM to make
it more human-adapted and to differentiate influences of trait anxiety and acrophobia by
introducing various floor textures and lower walls of closed arms to the height of standard
handrails. In the first experiment, hierarchical regression analyses of exploration behavior
revealed the expected association between open arm avoidance and Trait Anxiety, an
even stronger association with acrophobic fear. In the second study, results revealed that
acrophobia was associated with avoidance of open arms with mesh-floor texture, whereas
for trait anxiety, claustrophobia, and sensation seeking, no effect was detected. Also,
subjects’fear rating was moderated by all psychometrics but trait anxiety. In sum, both
studies consistently indicate that humans show no general open arm avoidance analogous
to rodents and that human EPM behavior is modulated strongest by acrophobic fear,
whereas trait anxiety plays a subordinate role. Thus, we conclude that the criteria for cross-
species validity are met insufficiently in this case. Despite the exploratory nature, our
studies provide in-depth insights into human exploration behavior on the virtual EPM.
Keywords: elevated plus-maze, EPM, anxiety, virtual reality, translational neuroscience, acrophobia, trait anxiety
INTRODUCTION
The Elevated Plus-Maze (EPM) is a widely used and well-accepted apparatus to measure anxiety in
rodents. Its advantage is that it assesses natural behavior and therefore does not require a previous
conditioning process or other preparations that might interfere with experimental variables
distorting behavioral outcomes. The EPM, with its open and closed arms, comprises a conflict
Edited by:
Barbara Rothbaum,
Emory University, United States
Reviewed by:
Mark Steven Burton,
Emory University, United States
Brian Dixon,
New Zealand Psychological Society,
New Zealand
Katarzyna Wyka,
City University of New York,
United States
*Correspondence:
Octavia Madeira
octavia.madeira@uni-wuerzburg.de
Specialty section:
This article was submitted to
Virtual Reality in Medicine,
a section of the journal
Frontiers in Virtual Reality
Received: 29 November 2020
Accepted: 18 February 2021
Published: 20 April 2021
Citation:
Madeira O, Gromer D, Latoschik ME
and Pauli P (2021) Effects of
Acrophobic Fear and Trait Anxiety on
Human Behavior in a Virtual
Elevated Plus-Maze.
Front. Virtual Real. 2:635048.
doi: 10.3389/frvir.2021.635048
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350481
ORIGINAL RESEARCH
published: 20 April 2021
doi: 10.3389/frvir.2021.635048
situation as rodents, on the one hand, prefer dark and enclosed
arms to avoid potential predators, but, on the other hand, are
naturally prone to engage actively in exploration behavior
towards novel stimuli (Barnett, 2017). Following-up a study of
Montgomery (1955),Pellow et al. (1985) developed an elevated
plus-shaped platform with two closed and two open arms. They
examined rats and observed that anxiolytic drugs (e.g., diazepam)
significantly increased open arm entries and open arm
exploration time, whereas anxiogenic medication induced
opposite effects.
These seminal findings strongly suggest that rodents’
movement patterns in the EPM are reliable and valid
indicators of general anxiety. Based on these findings and
numerous follow-up studies, the EPM is now considered a
standard test of anxiety in rodents and routinely used in
preclinical translational research, i.e., for drug or target gene
testing. For instance, the anxiolytic effect of benzodiazepines was
detected in rodents by using the EPM, and this is why diazepam is
now an established anxiolytic drug within the human clinical
setting until today.
One way to examine the translational validity of animal tests is
re-translation, i.e., the development of human analogs to animal
paradigms (Kirlic et al., 2017). Such re-translation proved to be
successful for several cognitive tasks, e.g., monetary- or
punishment-based conflict paradigms. First attempts to re-
translate animal paradigms assessing anxiety-related behavior
proved to be successful too (Kirlic et al., 2017). An advantage
of such re-translational human tasks is that they assess both
behaviors analogous to animal paradigms and verbal responses
which may help to validate the paradigm. For instance, Aupperle
et al. (2011) re-translated an approach-avoidance conflict task
frequently used in anxiety researches in animals to humans and
found that either conflict or reward approach is associated with
certain aspects of anxiety sensitivity including gender. Walz et al.
(2016) re-translated the open field test, which is frequently used
in rodents to measure anxiety. In rodents, increased thigmotaxis
during exploration of the open field, i.e., staying close to the open
field’s wall, is interpreted as an indicator of animal anxiety (Gould
et al., 2009). In analogy, Walz et al. asked their participants to
freely explore an open field, i.e., a soccer field surrounded by trees
and bushes (Walz et al., 2016). Results revealed that individuals
reporting agoraphobic fear or high levels of anxiety sensitivity, a
known risk factor for agoraphobia and panic disorder (Hofmann
et al., 2009), displayed an increased thigmotaxis, i.e., they moved
along the walls of the open field more and avoided its center for a
longer time. These results validate the open field animal studies
because they corroborate an association between reported level of
anxiety and open field behavior (Grillon and Ernst, 2016).
Interestingly, these findings also imply that characteristic
ethologic patterns in rodents are conserved in humans also
and seem to play a significant role in human anxiety.
However, there still exists a scientific gap regarding the cross-
species translational validity of these anxiety models. It is often
criticized that they lack predicational clinical value (Grillon and
Ernst, 2016). As a result, there is a paucity in developing new
anxiolytic agents or even therapeutic approaches based on these
preclinical “classic”tests (Griebel and Holmes, 2013;Grillon
et al., 2019). At the same time, the (re-) translation of
Pavlovian fear conditioning already provided numerous
insights to fear-related issues in humans while maintaining
methodological comparability and providing cross-species
validity for years by now (Lonsdorf et al., 2017;Haaker et al.,
2019). Consequently, closing this gap by testing cross-species
validity could similarly set new impulses in both human and
animal research.
Trait anxiety is an enduring and generalized predisposition to
react with fear in potentially threatening or ambiguous situations
(Spielberger, 1966;Spielberger, 2013). It was found that
individuals with high trait anxiety scores are prone to develop
an anxiety disorder over their lifetime (Hofmann et al., 2009;
Gallagher et al., 2014;Kindt and Soeter, 2014). On a behavioral
level, high trait anxiety manifests in harm avoidance,
i.e., avoidance behavior (Maner and Schmidt, 2006). From a
clinical point of view, findings on emotional and behavioral
mechanisms of subthreshold anxiety disorders (i.e., not yet
pathologic but close to meet diagnostic criteria) imply that it
can be similarly impairing as a fully developed pathologic
condition (Carter et al., 2001;Karsten et al., 2011). In light of
this scientific paucity, the human virtual EPM might be a
promising tool to investigate behavioral movement patterns
associated with trait anxiety and contribute to a translational
understanding of anxiety mechanisms. Moreover, the insights on
movement behavior related to trait anxiety might complement
current research methods such as self-report data and
physiological measures.
Virtual reality (VR) offers promising advancement in the
development of human analogs to animal paradigms. VR is a
computer-generated environment simulating sensory input,
mostly visual and auditory, in which the participant is
immersed, feels present, and can behave and interact with the
environment (see Sanchez-Vives and Slater, 2005, for a discussion
of its use in animal and human research; Bohil et al., 2011). For
instance, Dobricki and Pauli (2016) used virtual reality to assess
body-environment interactions and emotions in a virtual height
scenario in humans, and Gromer et al. (2019) systematically
examined the emotional responses of humans confronted with
virtual heights.
Biedermann et al. (2017) adopted a mixed virtual reality design
to develop a human elevated plus-maze (EPM) paradigm (see also
Madeira et al., 2017). Their virtual EPM consisted of a real-world
wooden maze providing haptic feedback combined with a virtual
maze providing visual feedback presented with a head-mounted
display (HMD). The maze was placed on a virtual rocky
mountain surrounded by water with two opposite arms, and
the center of the maze was surrounded by rocks while the other
two arms reached out over the water at 55 m height. Results
revealed that participants avoided open arms and reported more
anxiety on these arms. Significantly, open arm avoidance was
positively associated with acrophobic fear but not with anxious
temperament (STAI) or social anxiety and negatively associated
with sensation seeking. These opposing associations of
acrophobia and sensation seeking with open arm avoidance
behavior correspond with assumptions that both constructs
are related to opposing motivational systems (Boecker and
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350482
Madeira et al. Exploration in a Virtual EPM
Pauli, 2019). Finally, anxioselective drugs were found to decrease
and increase open arm avoidance as expected. While this study
has several strengths, it also has important limitations. First, walls
surround the closed arms of the rodent EPM, while the closed
arms of the designed human EPM had no walls. Instead, the
humans stood in both conditions on real wooden planks with
tangible borders that in one condition were perceived visually as
lying on rocky ground and in the other condition as high above
the water. Second, the used HMD caused a dissociation of haptic
from visual feedback as it precluded the perception of the own
body including the feet. In consequence, participants’movement
on the maze might have been unnatural requiring increased
attention on the feet and floor, and this might have
exaggerated the height perception and in consequence feelings
of insecurity and arousal especially on the “open”arms. Finally,
the rodent EPM behavior was found to be independent of height
(Treit et al., 1993) and is therefore considered as a measure of
general anxiety, e.g., it correlates with other anxiety measures like
the open field test or the dark-light box (Ramos et al., 2008). In
contrast, the observed human EPM behavior was related to fear of
height, but not trait anxiety. In sum, the author’s conclusion that
the “identical outcome parameters in the human and rodent EPM
facilitate translational research across species”(Biedermann et al.,
2017 p8) might be premature.
Here, we examined the construct validity of a virtual EPM
to assess the anxiety in humans in two studies. However, in
contrast to Biedermann et al. (2017), we used a 5-sided Cave
Automatic Virtual Environment (CAVE) instead of an HMD
to present the virtual environment with the advantage that it
allows participants to see the own body while simultaneously
being able to walk naturally. In our opinion, this better allows
free exploration comparable to a rodent EPM. Moreover, a
CAVE system allows a bigger field of view and triggers less
simulatorsicknessthananHMD(Rebenitsch and Owen,
2016), which might be a moderating variable for anxiety if
using VR scenarios with height situations. Also, we decided to
stick to the animal EPM design as close as possible and
maintain the experimental procedure. Thus, the typical
animal EPM (Komada et al., 2008)wasenlargedtohuman
proportions, and participants were asked to freely explore
this environment while their behavior was recorded. In
addition, we used ratings to assess anxiety at different
EPM locations, and questionnaires were used to evaluate
trait anxiety, acrophobic fear, and sensation seeking. Based
on the findings in the first study and on participants’
feedback, we decided to modify the virtual EPM in a
second study and re-tested our hypotheses. Assuming that
a human EPM allows behavioral assessment of anxiety in
humans in analogy to rodent studies, we hypothesized that 1)
humans exhibit a general open arm avoidance and that 2)
their level of trait anxiety is positively associated with open
arm avoidance and fear ratings on open arms. Alternatively,
we speculated following Biedermann et al. (2017) that 3)
acrophobic fear is positively associated with open arm
avoidances and fear ratings on open arms and that 4)
sensation seeking is negatively associated with open arm
avoidance.
STUDY 1
Methods
Sample
A total of 33 individuals, most of them students, were recruited
via email and flyers and received either 12 EUR or course credit
for participation. Due to technical problems during data
recording, three participants were excluded from analysis
resulting in a final sample of 30 individuals (10 males and 20
females) with a mean age of M21.93 (SD 2.85).
The study was carried out following the recommendations of the
Declaration of Helsinki and the Ethical Guidelines of the German
Psychological Society. The Ethics Committee of the Department of
Human-Computer-Interaction of the University of Würzburg
provided ethical approval for the protocol.
Apparatus
Assessments were conducted in the 5-sided 3D Multisensoric
PsyCave Laboratory of the Department of Psychology I of the
University of Würzburg sized 4 ×3×2.95 m. The visual
simulations on the five canvases (4 walls plus floor) were
presented via six projectors (NW-7, BARCO, Kuurne, Belgium)
with a resolution of 1920 ×1,200 pixels each and connected to two
computers per projector (Intel Core i7-2600K; 8GB RAM; Nvidia
Geforce GTX 580; OCZ Vertex2 SSDs). Participants wore passively
color-filtering glasses (Infitec Premium, Infitec, Ulm, Germany) for
stereoscopic effect. Auditory instructions were given via a 7.1
surround sound system (CANTON, Weilrod, Germany).
Motion tracking was recorded with four LED infrared cameras
installed on top of the CAVE canvases (PhaseSpace Impulse,
PhaseSpace Inc, San Leandro, CA, United States). A wireless
Xbox 360 controller (Microsoft, Redmond, WA, United States)
was used for navigation over long distances. Walking by foot was
possible within the space of the CAVE.
The experimental virtual environment was set up with a self-
made modification of the first-person shooter game, “Half Life 2:
Deathmatch”(Valve, Bellevue, Washington, United States) based
FIGURE 1 | Screenshot of the virtual Elevated Plus-Maze. Center area
marks the starting point of the exploration phase lasting for 5 min; starting
gaze direction was randomized across participants.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350483
Madeira et al. Exploration in a Virtual EPM
on Source Engine 2007. The experimental procedure was
managed by the CS-Research 5.6 software (VTplus, Würzburg,
Germany; see www.cybersession.info for detailed information)
(cf. Kinateder et al., 2014).
Virtual Scenario
The virtual human EPM was identical to the animal EPM but
adjusted to human proportions. Thus, the plus-shaped platform
had a height of 10 m with four maze arms arranged orthogonally
to each other, each 11 m long and 3 m wide. To control for
environmental cues that might interfere with exploration
behavior and to exacerbate height estimation for participants,
platform pillars were surrounded by fog. All arms had a solid
ground texture. The two open arms were free of any visible
barrier, whereas the two closed arms had wooden-textured walls
of 3 m in height (see Figure 1). Please note that we were unaware
of Biedermann et al.’s (2017) design as we conducted the study
prior to their publication (Madeira et al., 2017).
Tracking Data
The participants’positioning data within the virtual scenario
were tracked continuously with a sample rate of 60 Hz. Based on
this positioning data, we calculated the behavioral parameters
during the 5 min of exploration time: time spent on open and
closed arms, distance walked on open and closed arms, and total
walking distance. The center area was defined as a transition zone
and therefore excluded from analyses (Hogg, 1996).
Anxiety and Presence Ratings
After the free exploration and fading out of the virtual
environment, participants were teleported to the center area of
the maze again and asked to walk forward to a green footprint on
one of the arms and then rate the anxiety experienced on that
position with Subjective Units of Discomfort Scales (SUDS)
ranged from 0 to 100. Footprint markers were used to ensure
consistency in position for ratings across participants. These
ratings were assessed on one open and one closed arm
selected randomly. After the anxiety rating trial, the feeling of
presence was also assessed with the question “To which extent do
you feel present in the VE, i.e., as if you were really there?”on a
scale ranging from 0 to 100 (Bouchard et al., 2008).
Means and standard deviation of anxiety and presence ratings
are summarized in supplementary materials (Supplementary
Table S1).
Psychometric Questionnaires
The German version of the State-Trait-Anxiety Inventory (STAI;
Laux et al., 1981) is a self-report questionnaire assessing state
anxiety with 20 items (e.g., “I am nervous.”) rated on a scale from
1(“not at all”)to4(“very much”) and assessing trait anxiety with
20 items (e.g., “I am happy.”) rated on a scale from 1 (“almost
never”)to4(“almost always”). Sum scores range from 20 to 80.
The Acrophobia Questionnaire (AQ; Cohen, 1977) is a 40-item
self-report questionnaire measuring fear of heights (acrophobia)
on two subscales. The Anxiety subscale assesses fear towards
common height situations, e.g., take a ride on the Ferris wheel,
with seven-point Likert scales from 0 (“not anxious at all”)to7
(“extremely anxious”), resulting in a sum score range of 0–120. The
Avoidance subscale asks to rate the same items on a three-point
Likert scale from 0 (“would not avoid”)to3(“would not do it under
any circumstances”), thus assessing behavioral avoidance in
response to height situations with a sum score ranging from 0
to 40. Hüweler et al. (2009) and Diemer et al. (2016) already applied
the AQ to test the acrophobic symptoms in virtual environments.
The German version of the Sensation-Seeking Scale
(Beauducel et al., 2003) assesses the individual’s optimal levels
of stimulation and arousal with 40 items with a forced-choice
design. Participants are asked to choose one of two statements
that best reflect their personal beliefs (e.g., “I get bored seeing the
same old faces.”vs. “I like the comfortable familiarity of everyday
friends.”). The SSS-V has four subscales measuring “Thrill and
Adventure Seeking,”“Disinhibition,”“Experience Seeking,”and
“Boredom Susceptibility.”The total score ranging from 0 to 40
represents the sum score of all subscales. Due to technical
problems, 18 participants were unable to answer the last five
items of the SSS-V. To complete the data set, missing items were
estimated by means of other items on each subscale and then
rounded.
The Igroup Presence Questionnaire (IPQ; Schubert et al., 2001)
consists of 14 items and evaluates the feeling of presence within the
VR scenario as a self-report. Spatial Presence (e.g., “Somehow I felt,
that the virtual world surrounded me.”), Involvement (e.g., “Iwas
not aware of my real environment.”), and Experienced Realism
(e.g., “How real did the virtual world seem to you?”) are measured
on a seven-point Likert Scale. Additionally, a general item assesses
the “sense of being there.”All subscale scores range from 0 to 6. The
IPQ has been validated for various types of virtual environments
(see http://www.igroup.org/pq/ipq/index.php for further details).
Questionnaire scores (means and standard deviations) are
displayed in the supplementary materials section
(Supplementary Table S1).
Procedure
After signing the written informed consent, participants reported
sociodemographic data and then completed the questionnaires
(see above, except IPQ). Then, they entered the CAVE and
completed a virtual tutorial to adapt to navigation and audio
instructions.
The actual experiment started by teleporting participants to
the center area of the EPM with the instruction to explore the
environment for a total of 5 min freely. Starting gaze in the
direction of one of the arms was randomized across participants.
After 5 min, anxiety and presence were assessed on an open arm
and a closed arm as described above. Finally, participants
completed a questionnaire evaluating the sensation of presence
throughout the VR exposure (IPQ).
Statistical Analyses
Statistical analyses were performed with R and SPSS (RCoreTeam,
2013;IBM Corp., 2016). Exploration behavior was first analyzed
regarding differences between open vs. closed arms with dependent
samples t-tests. In addition, distance walked on the various maze
areas was analyzed via dependent samples t-test. Second,
associations between exploration behavior and psychometric
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350484
Madeira et al. Exploration in a Virtual EPM
questionnaires were analyzed with correlations; due to multiple
testing, Bonferroni corrections were applied. Third, hierarchical
regression analyses were performed to further evaluate the
influences on open arm activity by those psychometric
measures found to correlate with open arms avoidance. Finally,
the difference in anxiety ratings between arms was analyzed with a
t-test, and correlations between anxiety ratings on open and closed
arms with psychometric measures were performed.
Results
Exploration Behavior
Analyses of the motion tracking data revealed that participants
spent overall more time on the open (M151.52, SD 58.96)
compared to the closed (M60.54, SD 33.39) arms, t
(29)
6.36, p<0.001,d1.14 (see Figure 2). On average, subjects
walked a total of 191.72 m (SD 44.85) throughout the
exploration phase and 51.89 (SD 27.57) and 99.05 (SD
44.77) meters on the closed and open arms, respectively. The
difference of walking distance between the two arms was
significant, t
(29)
−4.52, p<0.001, d−0.83, indicating that
subjects were more active on the open than on the closed arms.
Correlational analyses (see Table 1) revealed that both trait
anxiety and acrophobia (both subscales) were negatively
associated with the time spent on the open arms. In contrast,
the SSS-V Total score was positively correlated with time spent on
open arms. However, after applying Bonferroni correction due to
multiple testing (p0.0063), only the negative correlation
between AQ Anxiety and time on open arms (r
(28)
−0.51,
p0.005) and the negative correlation between the SSS-V and
time spent on closed arms (r
(28)
−0.54,p0.002) remained
significant, suggesting that in humans, acrophobic fear is
associated with open arm avoidance, whereas proneness to
sensation seeking is linked to closed arm avoidance.
Correlation analyses of walking distances and psychometrics
revealed no effect (all pvalues >0.05).
A subsequent three-stage hierarchical multiple regression for
time spent on open arms as the dependent variable and trait
measures as predictors was conducted to evaluate the EPM
behavior further. Predictors were selected based on the results
of correlational analyses (see above), variables were added based
on our hypotheses and correlations with the dependent variable.
The test for multicollinearity indicated very low levels of
multicollinearity, and other assumptions for implementation of
regression models were not violated. In the first step, trait anxiety
was included to test for our main hypothesis. In a second step, the
Anxiety subscale of the AQ was added; the AQ Avoidance
subscale was deliberately excluded due to the strong
correlation between both measures (r
(27)
0.86, p<0.001)
suggesting multicollinearity. In the third step, the total score
of the SSS-V was included.
Comparison of all models via adjusted R
2
score (see Table 2)
indicated the second model to be best fit to explain time spent on
open arms, R
2
adj
0.28; F
2,27
6.307, p0.006. In this model, fear
of heights (AQ Anxiety) but not trait anxiety (STAI) significantly
FIGURE 2 | Heatmap of motion tracking data of study 1 using every fifth
recorded data point during the exploration phase (N30). Areas with lighter
colors depict areas with greater activity. The center area is defined as
transition zone, thus excluded from analyses.
TABLE 1 | Results of correlation analyses of time spent on open or closed arms
and psychometrics.
Time spent on open
arms
Time spent on closed
arms
rprp
STAI
Trait −0.38 0.043 0.17 0.371
AQ
Anxiety −0.51* 0.005 0.27 0.163
Avoidance −0.42 0.020 0.38 0.037
SSS-V
Total 0.36 0.048 −0.54* 0.002
Note: *Bonferroni-corrected to p <0.0063; AQ, Acrophobia Questionnaire; STAI, State-
Trait-Anxiety Inventory; SSS-V, Sensation Seeking Scale.
TABLE 2 | Summary of hierarchical regression analyses for psychometric
questionnaire scores predicting time spent on open arms of the virtual EPM in
study 1.
R
2
AIC B SE B βp
Step 1 0.13 319.96 0.033*
Intercept 264.64 53.03 <0.001**
STAI Trait −3.43 1.53 −0.40 0.033*
Step 2 0.28 303.40 0.006*
Intercept 258.70 48.24 <0.001**
STAI Trait −2.37 1.45 −0.28 0.114
AQ Anxiety −1.61 0.63 −0.43 0.017*
Step 3 0.26 305.29 0.018*
Intercept 250.91 75.32 0.003*
STAI Trait −2.36 1.48 −0.28 0.158
AQ Anxiety −1.54 0.80 −0.42 0.066
SSS-V Total 0.28 6.38 0.03 0.892
Note: *p <0.05, **p <0.01; R
2
adjusted R
2
; AQ, Acrophobia Questionnaire; STAI,
State-Trait-Anxiety Inventory; SSS-V, Sensation Seeking Scale.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350485
Madeira et al. Exploration in a Virtual EPM
explained exploration behavior suggesting that, in humans,
avoidance of open arms is primarily motivated by fear of heights.
Anxiety Ratings
T-test for dependent samples revealed significantly higher
anxietyratingonopen(M17.30, SD 20.38) vs. closed
arms (M4.77, SD 6.06), t
(29)
3.717, p<0.001. Further
analyses of correlations between psychometrically assessed
traits and anxiety ratings on open and closed arms revealed a
significant positive correlation (p-value was Bonferroni-
corrected to 0.0063) between anxiety rating on open arms
with the AQ Anxiety (r
(28)
0.53, p 0.003) and avoidance
subscales (r
(28)
0.62, p<0.001) (see Table 3). Thus, anxiety
ratings on open arms increased in line with acrophobic anxiety
and avoidance, whereas trait anxiety, including sensation
seeking, did not seem to be notably linked to anxiety ratings
on the maze.
Correlation of Trait Anxiety Measures
STAI-Trait and AQ subscale scores were uncorrelated (Anxiety:
r
(26)
0.29,p0.139; Avoidance: r
(27)
0.21, p 0.268), while
both AQ subscales were significantly negatively correlated with
the SSS-V score (all ps <0.05). Overview on questionnaire
intercorrelations is provided in supplementary materials
(Supplementary Table S2).
Summary
Contrary to our hypothesis, we found a general open arm
preference in this study. In line with our hypothesis, we found
that open arm avoidance is associated with trait anxiety.
However, our results revealed that avoidance of open arms
was best predicted by acrophobic fear. Additionally,
correlations between fear ratings on open arms and
acrophobic fear and acrophobic avoidance suggest that
participants with an increased trait of acrophobia experienced
more fear on the open arms and therefore exhibited increased
avoidance of them. The observed negative correlation between
acrophobic fear and sensation seeking may explain why we
also found some evidence that sensation seeking is associated
with decreased time spent on closed arms and less fear on open
arms, although not significant. These findings replicate the
associations found by Biedermann et al. (2017), who reported
similar relationships between open arm avoidance and
acrophobic fear including sensation seeking, which is
remarkable as the two virtual EPMs differed considerably in
design. These concordant results further validate that human
behavior in a virtual EPM measures anxiety, most likely
acrophobic fear.
STUDY 2
Based on the findings in study 1, we speculate that an association
between EPM behavior and trait anxiety might exist but was
overshadowed by a rather strong effect of acrophobic fear. Also,
we found no general open arm avoidance as observed in animal
studies. We think that this is due to the direct translation of the
rodent EPM with closed arms with rather high walls. This
restricted the view considerably and perhaps induced some
claustrophobic fear, as reported by some participants post-
experimentally, which might have reversed the effect and led
to closed arm avoidance. Therefore, we modified the virtual EPM
into a more human-adapted version and re-tested our
hypotheses.
Methods
Sample
The final sample of 61 participants (20 males, 41 females, mostly
students) had a mean age of M23.21 years (SD 3.74). Seven
examined participants were excluded due to technical problems
or premature completion of the experimental procedure.
Recruitment was performed via an online platform, flyers, and
emails. Participation was either monetary rewarded (12€)or
compensated with course credit.
Participants were selected via online screening. For this, we
extracted six items of the AQ and the Claustrophobia
Questionnaire (CLQ; Radomsky et al., 2001), respectively. To
assess subjects’level of acrophobia, they were asked to rate their
level of fear when confronted with six typical height situations
(corresponds with AQ Anxiety) on a five-point Likert scale and
the extent of avoidance behavior towards them (corresponds with
AQ avoidance) on a three-point Likert scale. For claustrophobic
fear, the six selected items were rated on a five-point Likert scale.
On a scale of sum scores ranging from 6 to 24 for acrophobic fear
(but not avoidance) including claustrophobia, only participants
scoring 12 or less, respectively, were invited. To examine
agoraphobic fear, we asked whether the subject ever
experienced a panic attack. If yes, we wanted to know whether
one or more of the listed situations, e.g., wide-open places, pulled
from the diagnostical criteria of the ICD-10 (Dilling and
Freyberger, 2012), are avoided out of fear of experiencing
another panic attack. Participants were excluded if
agoraphobic tendencies were present.
Additionally, estimation of the STAI Trait score was
conducted via four items with a four-point Likert scale but not
considered for inclusion.
TABLE 3 | Results of correlation analyses of anxiety ratings with psychometric
questionnaires.
Anxiety ratings on
open arms
Anxiety ratings on
closed arms
rprp
STAI
Trait 0.16 0.405 0.42 0.023
AQ
Anxiety 0.53* 0.003 0.32 0.095
Avoidance 0.62* <0.001 0.27 0.156
SSS-V
Total −0.45 0.013 −0.20 0.298
Note: Bonferroni-corrected to p <0.0063; AQ, Acrophobia Questionnaire; STAI, State-
Trait-Anxiety Inventory; SSS-V, Sensation Seeking Scale.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350486
Madeira et al. Exploration in a Virtual EPM
The study was carried out per the recommendations of the
Declaration of Helsinki and the Ethical Guidelines of the German
Psychological Society. The protocol was approved by the Ethics
Committee of the Institute of Human-Computer-Interaction. All
participants gave written informed consent.
Apparatus and Virtual Scenario
The apparatus was similar to that of study 1. The virtual EPM
scenario was modified as follows. Closed arms came with
handrails instead of walls with a height of 1.3 m, referring to
the official German standard for the height of handrails. The
arm’s width and length were reduced to 1.5 and 10 m,
respectively. Finally, we changed the arms floor with one open
and one closed arm having solid metal floors (as in study 1) and
one open and one closed arm having see-through metal grid
floors; the center area of the maze was divided diagonally and had
both surface textures (see Figure 3). Please note that we were
unaware of Biedermann et al.’s (2017) design as we conducted the
study prior to their publication (Madeira et al., 2017).
Tracking Data
Similar to study 1, positioning data within the virtual scenario
were tracked continuously with a sample rate of 60 Hz. For this
study, we calculated spent time and walked distance on each arm
individually to take the various arm types (open/closed) and floor
textures (solid/grid) into account. Again, we calculated the total
walked distance. The center area was defined as a transition zone
and therefore excluded from analyses (Hogg, 1996).
Procedure and Questionnaires
The experimental procedure was mostly similar to that of study 1,
with two exceptions. First, the Claustrophobia Questionnaire was
added and completed before the experiment started. Second,
anxiety ratings after the exploration trial were registered on all
four arms in a randomized sequence.
The German version of the Claustrophobia Questionnaire
(CLQ; Radomsky et al., 2001) evaluates fear of confined spaces
with overall 26 items assessing fear of suffocation (SS; 14 items,
e.g., “swimming with a nose clip”) and fear of restriction (RS; 12
items, e.g., “wear handcuffs for 15 min”) with five-point Likert
scales from 0 (“not anxious at all”)to4(“extremely anxious”).
Sum score of the SS scale ranges from 0 to 56, whereas sum score
of the RS vary between 0 and 48. The total questionnaire score
consists of the sum score of the two subscales and ranges from 0
to 104.
A summary of questionnaire scores, exploration, and rating
data are displayed in supplementary materials (see
Supplementary Table S3).
Statistical Analyses
First, differences between time spent and walking distances on
EPM arms were analyzed with a 2 ×2 repeated measure ANOVA
with the within-subject factor arm type (open vs. closed) and
floor texture (solid vs. grid). Second, associations between
exploration behavior and psychometric questionnaires were
analyzed with correlations; due to multiple testing, Bonferroni
corrections were applied. Third, hierarchical regression analyses
were performed to further evaluate influences on open arms
avoidance by those psychometric measures found to correlate
with open arms avoidance before. Finally, differences between
anxiety ratings on arms were analyzed with a 2 ×2 repeated
measures ANOVAs with the within-subjects factors arm type
(open vs. closed) and floor texture (solid vs. grid) and
correlations were calculated to analyze associations between
anxiety ratings on open and closed arms with psychometric
measures.
Results
Exploration Behavior
Analyses of the motion tracking data (see Figure 4B) with an
ANOVA with the within-subjects factors arm type (open vs.
closed) and floor texture (solid vs. grid) revealed no significant
main or interaction effects (all pvalues >0.05), hence indicating
no general preference of arm type or floor texture (see Figure 4).
Walking distance was averaged at 62.37 (SD 28.54) meters for
closed grid arm, 53.68 (SD 22.56) meters for closed solid, 44.47
(SD 24.12) meters for open grid, and 47.19 (SD 18.50) meters
for open solid arm, respectively. In addition, the mean of total
walking distance was M194.43 (SD 49.89). Statistical
comparison of walking distances with an ANOVA for within-
subject factor arm type and floor texture revealed a significant
main effect for arm type, F
1,60
4.384, p0.041, η
2
0.068, and a
significant interaction effect,F
1,60
5.621, p0.021, η
2
0.086.
No significant effect for floor type was found, F
1,60
0.759,
p0.387, η
2
0.012.Bonferroni-corrected post hoc t-tests
for paired samples (p0.008) revealed a significant difference
between walking distance on closed and open arm with solid
floor, t
(60)
−3.33, p0.002.
In sum, no general arm preference was observed, and walking
distances differed between open and closed arms with solid floor
but not between open and closed arms with grid floor or floor
textures in general.
Correlational analyses (see Table 4) revealed that increase in
AQ Anxiety and AQ Avoidance was associated with less time on the
open arm with grid floor (Anxiety: r
(59)
−0.36, p0.004;
FIGURE 3 | Screenshot of the virtual EPM used in the second study.
Walls on closed arms were lowered to handrail level and floor texture was
modified to consist of see-through mesh and solid metal floor.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350487
Madeira et al. Exploration in a Virtual EPM
Avoidance: r
(59)
-0.40, r0.001). Somewhat unexpected are the
observations that trait anxiety was associated with more time on the
open arm with grid floor (r
(59)
0.30, p0.021) and less time on the
closed arm with grid floor (r
(59)
−0.25, p0.05). No effect for
claustrophobic fear was detected. However, after Bonferroni
correction (p0.0025), only the negative correlation between
AQ Avoidance and time spent on the open grid floor arm
remained significant (p0.0013). These findings corroborate
study 1 and emphasize the importance of height perception for
human EPM behavior, whereas trait anxiety and claustrophobia
either play a subordinate or no role.
Moreover, a negative correlation (Bonferroni-corrected to p
0.003) was found between walking distance on open arm with
grid floor texture and AQ Avoidance, r
(59)
−0.346,p0.001.
A subsequent four-step hierarchical multiple regression
analysis was conducted to determine predictors for time spent
on open arms irrespective of floor texture, similar to study 1. Only
low levels of multicollinearity were present, and there was no
violation of relevant assumptions for regression analyses. The
predictors selected based on the correlational analyses and study
1findings were consecutively included in the model, i.e., trait anxiety,
fear of height (AQ Anxiety), and sensation seeking (SSS-V Total). To
estimate the influence of claustrophobia, we added the CLQ total
score in the fourth step.
The best-fitting model according to adjusted R
2
score for
predicting time spent on open arms included a linear
combination of STAI-Trait and AQ-Anxiety, R
2
adj
0.16,
F
2,58
5.35, p0.007 (Step 2). Within this model, AQ
Anxiety was the most important variable predicting the
avoidance of open arms (see Table 5). These results
replicate the findings of study 1 and also point to some
influence of trait anxiety.
Anxiety Ratings
A2×2 ANOVA on anxiety ratings with the within-subject factor
arm type (open vs. closed), and floor texture (grid vs. solid)
returned significant main effects of arm type, F
1,56
27.61, p<
0.001, η
2
0.33, and floor texture, F
1,56
12.33, p0.001, η
2
0.18, but no interaction effect, F
1,56
0.33,p0.569,η
2
0.01.
These findings implicate that participants rated open vs. closed
arms and grid floor vs. solid floor arms as more anxiety inducing.
Further correlation analyses revealed that the anxiety reported on
all EPM arms—except the “safest”with closed arm and solid
floor—were positively correlated (Bonferroni-corrected p0.0025)
FIGURE 4 | (A) Top view screenshot of the virtual EPM depicting the different floor textures; vertical arms are open, horizontal arms are closed. (B) Heatmap of
motion tracking data during the exploration phase (N61). Areas with lighter colors depict areas with greater activity.
TABLE 4 | Results of correlation analyses of time spent on various maze arms and psychometric questionnaires.
Closed grid Closed solid Open grid Open solid
rprp r prp
STAI
Trait −0.25 0.050 −0.10 0.463 0.29 0.021 0.08 0.548
AQ
Anxiety 0.02 0.863 0.15 0.253 −0.36 0.004 −0.12 0.377
Avoidance 0.04 0.768 −0.02 0.859 −0.40* 0.001 0.02 0.877
SSS-V
Total −0.19 0.150 −0.09 0.492 −0.13 0.328 0.06 0.656
CLQ
Total 0.04 0.773 −0.05 0.717 −0.08 0.539 −0.06 0.633
Note: *Bonferroni-corrected to p <0.0025; AQ, Acrophobia Questionnaire; STAI, State-Trait-Anxiety Inventory; SSS-V, Sensation Seeking Scale; CLQ, Claustrophobia Questionnaire.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350488
Madeira et al. Exploration in a Virtual EPM
with both AQ subscales (see Table 6). Furthermore, a significant
negative relationship was found between SSS-V Total score and all
arms except the open arm with a grid floor. Finally, significant positive
correlations were found between the CLQ total score and anxiety
ratings on all arms, whereas correlations between STAI-Trait and
anxiety ratings did not reach significance.
Correlations of Trait Anxiety Measures
Replicating study 1, we found again that AQ scores of both subscales
were uncorrelated with STAI-Trait (Anxiety: r
(59)
0.04,p0.715,
Avoidance: r
(59)
0.12, p0.347). Furthermore, we observed that
both AQ Anxiety and AQ Avoidance were positively associated
with the CLQ total score (Anxiety: r
(59)
0.39, p0.002;
Avoidance: r
(59)
0.43,p0.001). In addition, CLQ total score
and SSS-V-Total score correlated negatively, r
(59)
−0.28, p0.028
(see Supplementary Table S4 in supplementary materials). This
suggests that in this sample acrophobia is associated with
claustrophobia whereas sensation seeking is not or even
reversely linked to phobic anxiety.
Summary
With respect to the first hypothesis for this study, we again found no
general open arm avoidance in humans and no area preference in
general. Finally, the findings on fear ratings assessed on the various EPM
arms replicated study 1. Acrophobia but not trait anxiety positively
related to fear on all maze arms, except the least “acrophobic”one,
i.e., the closed arm with a solid floor. Also, sensation seeking was in
general negatively associated with fear ratings on EPM arms although
only the associations with ratings on the closed arm with grid floor and
open arm with solid floor survived significance. Unexpectedly, we
observed that claustrophobia was positively associated with fear ratings
on all areas of the maze, which may be due to the claustrophobic
situation of being in the CAVE system.
In sum, these results confirm the importance of acrophobia as
the driving factor for human EPM behavior, while trait anxiety
seems to play a rather subordinate role in open arm activity if any.
DISCUSSION
Inthetwostudies,weconceptuallytransferredtheEPM,awell-
established animal model of anxietymeasurement,toahumancontext
usingvirtualrealitytechnology.Equallytoanimalstudies(Pellow et al.,
1985), the participants’behaviors during 5 min of free exploration of a
virtual copy of an animal EPM adjusted to human proportions (study
1) or a human-adapted EPM (study 2) were registered. The construct
validity of human EPM behavior as a measure of anxiety was evaluated
by testing associations with psychometric measures of anxiety or
sensation seeking and fear ratings.
Exploration Behavior
Both of our studies observed no general open arm avoidance of
humans, which we expected based on animal studies and the
human EPM study of Biedermann et al. (2017). As a result, this
finding implies that exploration behavior on the animal and
human EPM is driven by mechanisms different from those in
the animal EPM. Unlike the human open field test, which
reproduced human thigmotaxis, we could not produce similar
evolutionary conserved behaviors (Walz et al., 2016). Therefore,
we conclude that we have no convincing evidence for cross-
species open arm avoidance suggested by Biedermann et al.
(2017).
TABLE 5 | Summary of hierarchical regression analyses for psychometric
questionnaire scores predicting time spent on open arms of the virtual maze in
study 2.
R
2
AIC B SE B βp
Step 1 0.04 616.50 0.122
Intercept 96.62 18.99 <0.001**
STAI Trait 0.77 0.49 0.20 0.122
Step 2 0.16 610.67 0.007**
Intercept 111.83 18.76 <0.001**
STAI Trait 0.83 0.46 0.22 0.079
AQ Anxiety −1.03 0.36 −0.34 0.007**
Step 3 0.11 612.56
Intercept 118.74 28.87 <0.001**
STAI Trait 0.80 0.47 0.21 0.095
AQ Anxiety −1.05 0.37 −0.35 0.007*
SSS-V Total −0.27 0.84 −0.04 0.753
Step 4 0.10 614.52
Intercept 116.93 30.90 <0.001**
STAI Trait 0.80 0.48 0.21 0.098
AQ Anxiety −1.07 0.40 −0.36 0.010*
SSS-V Total −0.23 0.87 −0.04 0.791
CLQ Total score 0.08 0.43 0.02 0.862
Note: *p <0.05, **p <0.01; R
2
adjusted R
2
; AQ, Acrophobia Questionnaire; STAI,
State-Trait-Anxiety Inventory; SSS-V, Sensation Seeking Scale.
TABLE 6 | Results of correlation analyses of anxiety ratings on various EPM arms with psychometric questionnaires in study 2.
Closed grid Closed solid Open grid Open solid
rprprprp
STAI
Trait 0.05 0.712 0.18 0.171 0.01 0.921 0.13 0.309
AQ
Anxiety 0.57* <0.001 0.36 0.004 0.49* <0.001 0.49* <0.001
Avoidance 0.46* <0.001 0.34 0.008 0.51* <0.001 0.50* <0.001
SSS-V
Total −0.43* <0.001 −0.38* 0.002 −0.38 0.003 −0.45* <0.001
CLQ
Total 0.45* <0.001 0.42* <0.001 0.43* <0.001 0.45* <0.001
Note: *p <0.0025. STAI, State-Trait-Anxiety Inventory; AQ, Acrophobia Questionnaire; SSS-V, Sensation Seeking Scale; CLQ, Claustrophobia Questionnaire.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 6350489
Madeira et al. Exploration in a Virtual EPM
Several differences between ours and the study of Biedermann
et al. (2017) might explain the discrepancy. At first, the definition
of the EPM zones differed from each other. The exploration
activity in the center area center activity remains a highly
discussed topic in animal research. It was found that it merely
associates with general locomotor activity but not anxiety (Braun
et al., 2011). Thus we decided not to incorporate center
movement into our final analyses. In addition, we followed the
advice of other authors to not merge maze areas for data analysis
(Hogg, 1996;Carobrez and Bertoglio, 2005;Braun et al., 2011).
Nonetheless, Biedermann et al. (2017) defined their maze’s closed
and center area as a “safe area”. Consequently, comparing a rather
large (safe) area to a much smaller area (open arms) likely leads to
an overestimation of time spent on these safe areas. In addition,
increased closed arm activity might also be the result of perceived
dangerousness in the context of the height situation. Considering
that human EPM behavior is mainly modulated by acrophobic
fear, it seems conceivable that the mixed virtual reality design
using a wooden cross with an arm width of 30 cm and a height of
20 cm amplifies fear in reaction to height cues, which prompted a
stronger avoidance of open arms than we observed in our studies.
Exploration Behavior and Anxiety
Both of our human EPM studies identified acrophobia to be the
dominant factor modulating open arm avoidance. Our results
corroborate Biedermann et al.’s(2017), who also reported
modulatory effects of acrophobia on open arm avoidance and
also added further insight. First, the modulatory effects of
acrophobia on human EPM behavior are observable for elevated
mazes of quite various designs, but visual height perception seems
to be a crucial factor in humans. Thus, the strongest associations
between acrophobia and open arm avoidance were observed in
EPMs with strong height simulation, i.e., in our study 2 with an
open arm with a see-through mesh ground and in Biedermann
et al. (2017) comparing elevated arms of 55 m height with safe
arms. A priori screening excluded participants with acrophobic
fear with the goal to unravel trait anxiety effects on EPM behavior.
Nonetheless, acrophobic fear was still the strongest modulator of
human EPM behavior as revealed by correlational and hierarchical
regression analyses. Second, our studies including Biedermann
et al. (2017) demonstrated that sensation seeking also has a
modulatory role for human EPM behavior. We observed that
an increase in sensation seeking is associated with less time on
closed arms. In contrast, Biedermann et al. (2017) reported that
subjects with higher sensation seeking levels tend to spend more
time on open arms. However, it remains questionable whether
these effects can be seen independent of acrophobia as both
questionnaires correlate negatively. Supporting this view, we
noticed that several items of the SSS-V (Zuckerman, Eysenck,
and Eysenck, 1978) refer to typical acrophobic situations (e.g., “I
like to dive off the high board.”vs “Idon’t like the feeling I get
standing on the high board (or I don’tgonearitatall).”). Further
studies have to evaluate whether acrophobia and sensation seeking
modulate human EPM behavior independently. Third, closed arms
with walls may be a safe area for rodents, but not for humans,
according to our studies. We speculate that humans feel confined
in such hallways and therefore are motivated to leave them. The
replacement of high solid walls by natural handrails allowing the
overview of the virtual environment may have reduced the
participants’motivation (e.g., due to claustrophobia or curiosity)
to leave the closed arms. Finally, we found some evidence for a
modulatory effect of trait anxiety on human EPM behavior.
However, these effects were weak and not consistent across the
two studies. This is in line with that of Biedermann et al. (2017)
who also found no modulatory role of trait anxiety. Newer studies
suggest that transdiagnostic concepts like fear of the unknown
(FOTU) and intolerance of uncertainty (IU) are underlying
mechanisms of trait anxiety, which are easier to target in an
experimental setting (Carleton et al., 2012;Carleton, 2016;
Shihata et al., 2016). Therefore, we suggest that future studies
introduce instruments to examine FOTU or IU (Carleton et al.,
2007) to gain further insights into nonphobic anxiety mechanisms
in a human EPM.To sum it up, we haveto conclude that, based on
the existing evidence, the human virtual EPM is instead a
behavioral test for acrophobia than for anxiety in general.
Anxiety Ratings
The anxiety ratings assessed on different EPM arms partly mirrored
the association of acrophobia and behavioral movement measures.
For example, participants reported enhanced anxiety on open vs.
closed arms and arms with see-through mesh floor vs. solid floor,
emphasizing the importance of visual height perception again. In
line, these ratings were associated with acrophobia but not trait
anxiety, and we found some negative associations between
sensation seeking and anxiety rating. In sum, these results
suggest that acrophobia and, to some extent sensation, seeking,
modulate human EPM behavior via the anxiety experienced on
different EPM arms. Interestingly, claustrophobic tendencies were
associated with anxiety ratings on all arms. Therefore, we speculate
that the five-sided CAVE system is a claustrophobic context
intrinsically, as we did not observe a consistent distinction
between arm types or floor texture in the second study.
Consequently, future studies have to consider and evaluate how
the technology used to present virtual environments moderates the
experienced anxiety and perhaps human EPM behavior.
Limitations
Our studies assessed behavioral and verbal measures of anxiety but
lack psychophysiological measures. Future studies should consider
psychophysiological measures as they may indicate anxiety responses
independent of self-report or behavioral measures (Lang, 1985). In
addition, more detailed analyses of movement patterns should be
considered. Besides exploration time as assessed here, rodent EPM
studies also assess other fear-like responses, such as freezing. In
humans, freezing is defined as reduced mobility and bradycardia
and has already been observed in specific human experimental settings
(Roelofs et al., 2010;Hagenaars et al., 2014;Rösler and Gamer, 2019).
Assessing freezing as human EPM behavior would be especially
valuable as participants may enter open arm alleys but abruptly
freeze, increasing the time spent on that specificmazearealeading
to erroneous interpretation of results.
Besides, the generalization of these findings might be limited
by the small age range within the sample consisting
predominantly of students.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 63504810
Madeira et al. Exploration in a Virtual EPM
Finally, several participants reported that the virtual
environments appeared somewhat artificial. Thus, future
studies should think about naturalistic EPM environments to
increase the test’s validity.
CONCLUSION
Our studies show that human movement in a virtual human EPM is
primarily influenced by acrophobia but not by trait anxiety, as we
initially hypothesized. The results are even more interesting, as, in
the second study, participants were selected for low levels of
acrophobia to avoid distortion of results. To address the
influence of trait anxiety on exploration behavior, it is essential to
reshape the virtual EPM in future studies to avoid the influence of
fear of heights. To target trait anxiety-related processes, we suggest
the subliminal induction of a threat to trigger a generalized fear state
typically associated with high trait anxiety (e.g., Van Den Hout et al.,
1995;Li et al., 2007;Reuman et al., 2015). This can either be achieved
by a threat of shock paradigm (Clark et al., 2012;Morriss et al., 2019)
or by emotional priming (Yang et al., 2016).
The present studies examined the influence of anxiety and
personality traits on human exploration behavior in detail. In
conclusion, we consider the virtual EPM an essential apparatus in
the investigation of anxiety-related issues as the virtual scenario
can easily be modified.
DATA AVAILABILITY STATEMENT
The original contributions presented in the studies can be found
on https://osf.io/j2fsw/. Further inquiries can be directed to the
corresponding author.
ETHICS STATEMENT
The studies involving human participants were reviewed and
approved by the Ethical Board of the Department of Human-
Computer-Interaction of the University of Würzburg. All
participants provided their written informed consent to
participate in this study.
AUTHOR CONTRIBUTIONS
OM, DG, MEL, and PP contributed to the study concept and
design. Data collection was done by OM. OM and DG performed
data analysis and interpretation. OM and PP drafted the
manuscript, and all other co-authors provided critical revisions
and feedback. All authors contributed to the article and approved
the submitted version.
FUNDING
The studies were supported by the Volkswagen Foundation
(AZ 94 102). This work was carried out with the support of
the structural doctoral program “Graduate School of Life
Sciences (GSLS)”by the University of Würzburg, and the two
studies are part of the doctoral project “The human-
experimental virtual Elevated Plus-Maze as an anxiety
model”ofOM.Thispublicationwassupportedbythe
Open Access Publication Fund of the University of
Wuerzburg.
ACKNOWLEDGMENTS
We would like to thank Jan Philipp Gast for assisting in data
collection and preparation.
SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at:
https://www.frontiersin.org/articles/10.3389/frvir.2021.635048/
full#supplementary-material.
REFERENCES
Aupperle,R.L.,Sullivan,S.,Melrose,A.J.,Paulus,M.P.,andStein,M.B.
(2011). A reverse translational approach to quantify approach-avoidance
conflict in humans. Behav. Brain Res. 225 (2), 455–463. doi:10.1016/j.bbr.
2011.08.003
Barnett, S. A. (2017). The rat: a study in behavior. London, United Kingdom:
Routledge.
Beauducel, A., Strobel, A., and Brocke, B. (2003). Psychometrische eigenschaften
und normen einer deutschsprachigen fassung der sensation seeking-skalen,
form V. Diagnostica 49 (2), 61–72. doi:10.1026/0012-1924.49.2.61
Biedermann, S. V., Biedermann, D. G., Wenzlaff, F., Kurjak, T., Nouri, S., Auer,
M. K., et al. (2017). An elevated plus-maze in mixed reality for studying
human anxiety-related behavior. BMC Biol. 15 (1), 125. doi:10.1186/s12915-
017-0463-6
Boecker, L., and Pauli, P. (2019). Affective startle modulation and psychopathology:
implications for a ppetitive and defensive brain systems. Neurosci. Biobehavioral
Rev. 103, 230–266. doi:10.1016/j.neubiorev.2019.05.019
Bohil, C. J., Alicea, B., and Biocca, F. A. (2011). Virtual reality in neuroscience
research and therapy. Nat. Rev. Neurosci. 12 (12), 752–762. doi:10.1038/
nrn3122
Bouchard, S., St-Jacques, J., Robillard, G., and Renaud, P. (2008). Anxiety increases
the feeling of presence in virtual reality. Teleoperators Virtual Environ. 17 (4),
376–391. doi:10.1162/pres.17.4.376
Braun, A. A., Skelton, M.R., Vorhees, C. V., and Williams, M. T.(2011). Comparison
of the elevated plus and elevated zero mazes in treated and untreated male
Sprague-Dawley rats: effects of anxiolytic and anxiogenic agents. Pharmacol.
Biochem. Behav. 97 (3), 406–415. doi:10.1016/j.pbb.2010.09.013
Carleton, N. R. (2016). Fear of the unknown: one fear to rule them all?. J. Anxiety
Disord. 41, 5–21. doi:10.1016/j.janxdis.2016.03.011
Carleton, R. N., Peter, M. A., Norton, J., and Asmundson, G. J. (2007). Fearing the
unknown: a short version of the intolerance of uncertainty scale. J. Anxiety
Disord. 21 (1), 105–117. doi:10.1016/j.janxdis.2006.03.014
Carleton, R. N., Mulvogue, M. K., Thibodeau, M. A., McCabe, R. E., Antony, M. M.,
and Asmundson, G. J. (2012). Increasingly certain about uncertainty:
intolerance of uncertainty across anxiety and depression. J. Anxiety Disord.
26 (3), 468–479. doi:10.1016/j.janxdis.2012.01.011
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 63504811
Madeira et al. Exploration in a Virtual EPM
Carobrez, A. P., and Bertoglio, L. J. (2005). Ethological and temporal analyses of
anxiety-like behavior: the elevated plus-maze model 20 years on. Neurosci.
Biobehav Rev. 29 (8), 1193–1205. doi:10.1016/j.neubiorev.2005.04.017
Carter, R. M., Wittchen, H. U., Pfister, H., and Kessler, R. C. (2001). One-year
prevalence of subthreshold and threshold DSM-IV generalized anxiety disorder
in a nationally representative sample. Depress. Anxiety 13 (2), 78–88. doi:10.
1002/da.1020
Clark,L.,Li,R.,Wright,C.M.,Rome,F.,Fairchild,G.,Dunn,B.D.,etal.
(2012). Risk-avoidant decision making increased by threat of electric shock.
Psychophysiology 49 (10), 1436–1443. doi:10.1111/j.1469-8986.2012.
01454.x
Cohen, D. C. (1977). Comparison of self-report and overt-behavioral procedures
for assessing acrophobia. Behav. Ther. 8 (1), 17–23. doi:10.1016/S0005-
7894(77)80116-0
Diemer, J., Lohkamp, N., Mühlberger, A., and Zwanzger, P. (2016). Fear and
physiological arousal during a virtual height challenge--effects in patients with
acrophobia and healthy controls. J. Anxiety Disord. 37, 30–39. doi:10.1016/j.
janxdis.2015.10.007
Dilling, H., and Freyberger, H. J. (2012). Taschenführer zur ICD-10-Klassifikation
psychischer Störungen. Berlin, Heidelberg: Springer.
Dobricki, M., and Pauli, P. (2016). Sensorimotor body-environment interaction
serves to regulate emotional experience and exploratory behavior. Heliyon 2
(10), e00173. doi:10.1016/j.heliyon.2016.e00173
Gallagher, M. W., Bentley, K. H., and Barlow, D. H. (2014). Perceived control and
vulnerability to anxiety disorders: a meta-analytic review. Cogn. Ther. Res. 38
(6), 571–584. doi:10.1007/s10608-014-9624-x
Gould, T. D., Dao, D. T., and Kovacsics, C. E. (2009). “The open field test,”in Mood
and anxiety related phenotypes in mice: characterization using behavioral tests.
Editor T. D. Gould (Totowa, NJ: Humana Press), 1–20.
Griebel, G., and Holmes, A. (2013). 50 years of hurdles and hope in anxiolytic drug
discovery. Nat. Rev. Drug Discov. 12 (9), 667–687. doi:10.1038/nrd4075
Grillon, C., and Ernst, M. (2016). Gain in translation: is it time for thigmotaxis
studies in humans?. Biol. Psychiatry 80 (5), 343–344. doi:10.1016/j.biopsych.
2016.07.003
Grillon, C., Robinson, O. J., Cornwell, B., and Ernst, M. (2019). Modeling anxiety
in healthy humans: a key intermediate bridge between basic and clinical
sciences. Neuropsychopharmacol. 44 (12), 1999–2010. doi:10.1038/s41386-
019-0445-1
Gromer, D., Reinke, M., Christner, I., and Pauli, P. (2019). Causal interactive links
between presence and fear in virtual reality height exposure. Front. Psychol. 10
(141), 141. doi:10.3389/fpsyg.2019.00141
Haaker, J., Maren, S., Andreatta, M., Merz, C. J., Richter, J., Richter, S. H., et al.
(2019). Making translation work: harmonizing cross-species methodology in
the behavioural neuroscience of Pavlovian fear conditioning. Neurosci.
Biobehavioral Rev. 107, 329–345. doi:10.1016/j.neubiorev.2019.09.020
Hagenaars, M. A., Oitzl, M., and Roelofs, K. (2014). Updating freeze: aligning
animal and human research. Neurosci. Biobehav Rev. 47, 165–176. doi:10.1016/
j.neubiorev.2014.07.021
Hofmann, S. G., Alpers, G. W., and Paul, P. (2009). “Phenomenology of panic and
phobic disorders,”in Oxford library of psychology. Oxford handbook of anxiety
and related disorders. Editors M. M. Antony and M. B. Stein (New York, NY:
Oxford University Press), 34–46.
Hogg, S. (1996). A review of the validity and variability of the elevated plus-maze as
an animal model of anxiety. Pharmacol. Biochem. Behav. 54 (1), 21–30. doi:10.
1016/0091-3057(95)02126-4
Hüweler, R., Kandil, F. I., Alpers, G. W., and Gerlach, A. L. (2009). The impact of
visual flow stimulation on anxiety, dizziness, and body sway in individuals with
and without fear of heights. Behav. Res. Ther. 47 (4), 345–352. doi:10.1016/j.
brat.2009.01.011
IBM Corp. (2016). IBM SPSS statistics for windows, version 24.0. Armonk, NY: IBM
Corp.
Karsten, J., Nolen, W. A., Penninx, B. W., and Hartman, C. A. (2011). Subthreshold
anxiety better defined by symptom self-report than by diagnostic interview.
J. Affect Disord. 129 (1), 236–243. doi:10.1016/j.jad.2010.09.006
Kinateder, M., Ronchi, E., Gromer, D., Müller, M., Jost, M., Nehfischer, M., et al.
(2014). Social influence on route choice in a virtual reality tunnel fire. Transp.
Res. F Traffic Psychol. Behav. 26, 116–125. doi:10.1016/j.trf.2014.06.003
Kindt, M., and Soeter, M. (2014). Fear inhibition in high trait anxiety. PLoS One 9
(1), e86462. doi:10.1371/journal.pone.0086462
Kirlic, N., Young, J., and Aupperle, R. L. (2017). Animal to human translational
paradigms relevant for approach avoidance conflict decision making. Behav.
Res. Ther. 96, 14–29. doi:10.1016/j.brat.2017.04.010
Komada, M., Takao, K., and Miyakawa, T. (2008). Elevated plus maze for mice.
J. Vis. Exp. 22, e1088. doi:10.3791/1088
Lang, P. J. (1985). “The cognitive psychophysiology of emotion: fear and anxiety,”
in Anxiety and the anxiety disorders. Editors A. Hussain Tuma and J. D. Maser
(Hillsdale, NJ: Lawrence Erlbaum Associates), 131–170.
Laux, L., Glanzmann, P., Schaffner, P., and Spielberger, C. D. (1981). State-trait
anxiety inventory—manual of the German version. Weinheim, Germany: Beltz.
Li, W., Zinbarg, R. E., and Paller, K. A. (2007). Trait anxiety modulates
supraliminal and subliminal threat: brain potential evidence for early and
late processing influences. Cogn. Affect Behav. Neurosci. 7(1),25–36. doi:10.
3758/cabn.7.1.25
Lonsdorf, T. B., Menz, M. M., Andreatta, M., Fullana, M. A., Golkar, A., Haaker, J.,
et al. (2017). Don’t fear “fear conditioning”: methodological considerations for
the design and analysis of studies on human fear acquisition, extinction, and
return of fear. Neurosci. Biobehavioral Rev. 77, 247–285. doi:10.1016/j.
neubiorev.2017.02.026
Madeira, O., Gromer, D., and Paul, P. (2017). “Transferring the elevated plus-maze
to a human context: a virtual reality study,”in Journal of neural transmission.
(Wien, Austria: Springer Wien), 1309.
Maner, J. K., and Schmidt, N. B. (2006). The role of risk avoidance in anxiety.
Behav. Ther. 37 (2), 181–189. doi:10.1016/j.beth.2005.11.003
Montgomery, K. C. (1955). The relation between fear induced by novel stimulation
and exploratory behavior. J. Comp. Physiol. Psychol. 48 (4), 254–260. doi:10.
1037/h0043788
Morriss, J., Saldarini, F., Chapman, C., Pollard, M., and van Reekum, C. M. (2019).
Out with the old and in with the new: the role of intolerance of uncertainty in
reversal of threat and safety. J. Exp. Psychopathol. 10 (1), 204380871983445.
doi:10.1177/2043808719834451
Pellow, S., Chopin, P., File, S. E., and Briley, M. (1985). Validation of open:closed
arm entries in an elevated plus-maze as a measure of anxiety in the rat.
J. Neurosci. Methods 14 (3), 149–167. doi:10.1016/0165-0270(85)90031-7
R Core Team (2013). R: a language and environment for statistical computing.
Vienna, Austria: R Core Team.
Radomsky, A. S., Rachman, S., Thordarson, D. S., McIsaac, H. K., and Teachman,
B. A. (2001). The claustrophobia questionnaire. J. Anxiety Disord. 15 (4),
287–297. doi:10.1016/S0887-6185(01)00064-0
Ramos, A., Pereira, E., Martins, G. C., Wehrmeister, T. D., and Izídio, G. S. (2008).
Integrating the open field, elevated plus maze and light/dark box to assess
different types of emotional behaviors in one single trial. Behav. Brain Res. 193
(2), 277–288. doi:10.1016/j.bbr.2008.06.007
Rebenitsch, L., and Owen, C. (2016). Review on cybersickness in applications and
visual displays. Virtual Reality 20 (2), 101–125. doi:10.1007/s10055-016-0285-9
Reuman, L., Jacoby, R. J., Fabricant, L. E., Herring, B., and Abramowitz, J. S. (2015).
Uncertainty as an anxiety cue at high and low levels of threat. J. Behav. Ther.
Exp. Psychiatry 47, 111–119. doi:10.1016/j.jbtep.2014.12.002
Roelofs, K., Hagenaars, M. A., and Stins, J. (2010). Facing freeze: social threat
induces bodily freeze in humans. Psychol. Sci. 21 (11), 1575–1581. doi:10.1177/
0956797610384746
Rösler, L., and Gamer, M. (2019). Freezing of gaze during action preparation under
threat imminence. Sci. Rep. 9 (1), 17215. doi:10.1038/s41598-019-53683-4
Sanchez-Vives, M. V., and Slater., M. (2005). From presence to consciousness
through virtual reality. Nat. Rev. Neurosci. 6 (4), 332–339. doi:10.1038/nrn1651
Schubert, T., Friedmann, F., and Regenbrecht, H. (2001). Igroup presence questionnaire.
Teleoperators Virtual Environ. 41, 115–124. doi:10.1162/105474601300343603
Shihata, S., McEvoy, P. M., Mullan, B. A., and Carleton, R. (2016). Intolerance of
uncertainty in emotional disorders: what uncertainties remain?. J. Anxiety
Disord. 41, 115–124. doi:10.1016/j.janxdis.2016.05.001
Spielberger, C. D. (2013). Anxiety: current trends in theory and research.
Amsterdam, Netherlands: Elsevier.
Spielberger, C. D. (1966). Anxiety Behavior. Cambridge, MA: Academic Press.
Treit, D., Menard, J., and Royan, C. (1993). Anxiogenic stimuli in the elevated plus-maze.
Pharmacol. Biochem. Behav. 44 (2), 463–469. doi:10.1016/0091-3057(93)90492-c
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 63504812
Madeira et al. Exploration in a Virtual EPM
Van Den Hout,M., Tenney, N., Huygens, K., Merckelbach, H., and Kindt, M. (1995).
Responding to subliminal threat cues is related to trait anxiety and emotional
vulnerability: a successful replication of MacLeod and Hagan (1992). Behav. Res.
Ther. 33 (4), 451–454. doi:10.1016/0005-7967(94)00062-o
Walz, N., Mühlberger, A., and Pauli, P. (2016). A human open field test reveals
thigmotaxis related to agoraphobic fear. Biol. Psychiatry 80 (5), 390–397. doi:10.
1016/j.biopsych.2015.12.016
Yang, Y., Lueken, U., Wittmann, A., Holtz, K., Kleint, N. I., Herrmann, M. J., et al.
(2016). Neural correlates of individual differences in anxiety sensitivity: an
fMRI study using semantic priming. Soc. Cogn. Affect. Neurosci. 11 (8),
1245–1254. doi:10.1093/scan/nsw024
Zuckerman, M., Eysenck, S., and Eysenck, H. J. (1978). Sensation seeking in
England and America: cross-cultural, age, and sex comparisons. J. Consult Clin.
Psychol. 46 (1), 139–149. doi:10.1037//0022-006x.46.1.139
Conflict of Interest: PP is a shareholder of a commercial company that develops
virtual environment research systems (VTplus GmbH) for empirical studies in the
field of psychology, psychiatry, and psychotherapy.
The remaining authors declare that the research was conducted in the absence of
any commercial or financial relationships that could be construed as a potential
conflict of interest.
Copyright © 2021 Madeira, Gromer, Latoschik and Pauli. This is an open-access
article distributed under the terms of the Creative Commons Attribution License (CC
BY). The use, distribution or reproduction in other forums is permitted, provided the
original author(s) and the copyright owner(s) are credited and that the original
publication in this journal is cited, in accordance with accepted academic practice. No
use, distribution or reproduction is permitted which does not comply with these terms.
Frontiers in Virtual Reality | www.frontiersin.org April 2021 | Volume 2 | Article 63504813
Madeira et al. Exploration in a Virtual EPM
