Content uploaded by Deltcho Valtchanov
Author content
All content in this area was uploaded by Deltcho Valtchanov on May 21, 2014
Content may be subject to copyright.
Content uploaded by Deltcho Valtchanov
Author content
All content in this area was uploaded by Deltcho Valtchanov on Mar 05, 2014
Content may be subject to copyright.
Restorative Effects of Virtual Nature Settings
Deltcho Valtchanov, B.Sc., Kevin R. Barton, M.A., and Colin Ellard, Ph.D.
Abstract
Previous research regarding the potential benefits of exposing individuals to surrogate nature (photographs and
videos) has found that such immersion results in restorative effects such as increased positive affect, decreased
negative affect, and decreased stress. In the current experiment, we examined whether immersion in a virtual
computer-generated nature setting could produce restorative effects. Twenty-two participants were equally di-
vided between two conditions, while controlling for gender. In each condition, participants performed a stress-
induction task, and were then immersed in virtual reality (VR) for 10 minutes. The control condition featured
a slide show in VR, and the nature experimental condition featured an active exploration of a virtual forest.
Participants in the nature condition were found to exhibit increased positive affect and decreased stress after
immersion in VR when compared to those in the control condition. The results suggest that immersion in virtual
nature settings has similar beneficial effects as exposure to surrogate nature. These results also suggest that VR
can be used as a tool to study and understand restorative effects.
Introduction
Restorative effects are classified as a reduction in
cognitive fatigue, decreased stress levels, increased focus,
increased positive affect, decreased negative affect, and de-
creased sympathetic nervous system activity.
1,2
Early studies
by Kaplan and Kaplan
3,4
measured cognitive and emotional
changes in participants in an outdoor challenge program. The
program consisted of instruction in survival skills for a pe-
riod of 2 weeks. Pre- and posttests and comparisons with
control groups revealed gains in confidence, self-sufficiency,
and more realistic self-evaluations. More recent research by
Berman et al.
5
compared effects of interaction with natural
versus urban environments on cognitive function. Berman
et al.
5
found that a walk in a local park allowed for greater
improvement in performance on a backwards digit-span task
than a walk through a downtown area. Berman et al.
5
also
found that executive attention (as measured by the Attention
Network Task) and performance on a backwards digit-span
task improved after viewing pictures of nature when com-
pared to a group that viewed urban areas.
Findings such as those of the Kaplans gave rise to the at-
tention restoration theory (ART) as described in Kaplan,
6,7
stating that when one interacts with an environment (such as
a natural environment) that is rich with fascinating stimuli,
attention is modestly captured in a bottom-up fashion by
the stimuli, allowing directed attention mechanisms to re-
cover.
2,6,7
Also, unlike natural environments that modestly
promote involuntary attention, urban environments contain
bottom-up stimuli (such as flashing neon signs, lights, and
various loud sounds) that dramatically capture attention while
also requiring directed attention to overcome the stimuli (e.g.,
ignoring advertisements), and are thus less restorative.
5
Re-
cent research by Berto et al.
8
examining eye movements when
looking at images of urban and nature settings demonstrated
fewer fixations in high-fascination (nature) scenes, suggesting
that they were viewed with less effort and providing support
for the notion that nature modestly promotes involuntary
attention as suggested by ART.
In contrast to Kaplan’s theory, Ulrich et al.
9,10
proposed
that an individual’s initial response to an environment is af-
fective rather than cognitive. In this theory, structural prop-
erties (such as complexity and focality) are believed to elicit
an automatic affective response. Ulrich proposed that the
patterns within a person’s visual field (such as those in nature
settings) prompt prepotent (and potentially automatic) re-
sponses from the individual.
2,9
In this theory, patterns seen
outside of nature are believed to be more threatening (and
physiologically arousing) than those seen in nature settings.
2
Unlike Kaplan’s model, where replenishment of capacity for
attention is believed to be the source of restoration, Ulrich’s
model proposes that the initial affective response (including
that of liking and moderate interest) to an environment
shapes cognitive events that follow, leading to sustained at-
tention, higher levels of positive feelings, and reduced nega-
tive or stress-related feelings, as well as suppressed negative
thoughts.
10
Experiments by Ulrich have focused on measur-
ing the emotional and physiological states of participants.
Department of Psychology, University of Waterloo, Ontario, Canada.
CYBERPSYCHOLOGY,BEHAVIOR, AND SOCIAL NETWORKING
Volume 13, Number 5, 2010
ªMary Ann Liebert, Inc.
DOI: 10.1089=cyber.2009.0308
503
In one experiment, participants were shown slides of either
a natural scene with vegetation or an urban setting, while
measurements of heart rate and alpha amplitude were taken.
9
Later research by Ulrich has shown that videos of nature can
reduce heart rate and skin-conductance level (SCL),
10
and
promote a trend decrease in blood pressure.
11
More recent
research by Ulrich also demonstrates that even posters of
nature can reduce stress in the workplace.
12
Results of his
research are consistent with those of Kaplan and Kaplan
1,2
and Berman et al.
5
in demonstrating the power of nature
environments as restorative settings. More specifically, Ulrich
et al.
9,10
demonstrated that even surrogate nature (in the form
of photographs and videos) can have a significant effect on
the physiological, emotional, and cognitive states of indi-
viduals.
Since the emergence of these two theories, the questions of
‘‘how’’ and ‘‘why’’ restorative environments reduce cognitive
fatigue, decrease stress levels, and increase an individual’s
ability to focus have not been thoroughly researched. Instead,
there has been a shift from theory to application of restorative
environments in increasing people’s quality of life. This in-
cludes things such as increasing job satisfaction and well-
being in the workplace by adding windows that overlook
natural scenery and increase natural light.
13
A recent study by
Weinstein et al.
14
examined how the presentation of nature
versus urban images influenced participants’ aspirations and
feeling of autonomy and generosity. The study found that
when exposed to nature, individuals valued intrinsic aspira-
tions more (such as personal growth, intimacy, and connect-
ing with others) and devalued extrinsic aspirations (such as
material wealth, personal image, and fame).
14
The results also
suggested that participants’ feeling of autonomy and gener-
osity increased after exposure to nature.
14
This line of re-
search converges with findings by Ulrich,
9
Ulrich et al.,
10,11
Kweon et al.,
12
Kaplan,
3,4
and Berman et al.
5
on the benefi-
cial aspects of natural environments, and further expands on
previous research to suggest that nature may not only reduce
physiological and cognitive levels of stress, but may also
promote a healthier psychological mindset that values in-
trinsic aspirations, autonomy, and generosity. This converg-
ing evidence on the beneficial effects of nature prompts an
important question: can restorative environments be created
and customized to promote a healthy psychological mindset,
and help individuals deal with different types of stress?
Modern advancements in immersive virtual reality (VR)
technology would allow for the complete customization and
creation of restorative environments. Previous work, such as
that by Ulrich et al.,
10,11
van den Berg et al.,
15
and de Kort
et al.,
16
has shown that the restorative effects described by
Kaplan
6,7
and Ulrich
9
can be elicited by ‘‘virtual nature,’’ by
using films of real nature displayed on screens. Here, we
would like to make the distinction between ‘‘virtual nature’’
as described by de Kort et al.,
16,17
which encompasses films of
real nature presented on displays, and the computer-generated
nature presented in immersive VR. Computer-generated na-
ture consists of an artist’s interpretation and rendition of what
natural objects, colours, shapes, sounds, physics, motion, and
light look like. Also, unlike real nature, computer-generated
nature is created with a finite number of variables (objects,
shapes, colours, physics, etc) defined by the researcher. This
level of control results in differences between real and virtual
nature. Due to the finite number of variables, plants and trees
in virtual nature are much more similar to each other than
those found in real nature. When considering all of these
factors, computer-generated nature is better perceived as an
interpretation of what nature is, rather than a replication (such
as a photograph or video of real nature). Lastly, the interac-
tion with computer-generated nature in VR as presented in
this study allows for free exploration of the synthetic envi-
ronment, whereas the use of film and screens does not have
such affordances. Thus it should be noted that the computer-
generated nature in the current study is not a direct replication
(in the form of video or photograph) like in previous studies.
In the current study, it was hypothesized that exposure to
natural environments in VR would significantly reduce
stress, cognitive fatigue, and negative affect, while increasing
positive affect in participants who had recently experienced a
stressful event. Several measures were employed in order to
quantify stress, cognitive fatigue, and affect: SCL and heart
rate in the form of the interbeat interval (IBI), both of which
have been used as measures of physiological state and stress
levels,
10,16,18
in combination with affective measures such as
the Zuckerman Inventory of Personal Reactions (ZIPERS).
10
Skin conductance has been developed as a direct physio-
logical measure of sympathetic nervous system activity.
21
Increases in sympathetic nervous system activity are accom-
panied by increases in activation of epidermal tissue (in-
cluding sweat glands) and secretion of sweat, which results in
increased skin conductivity.
21
Skin-conductance measures
contain two different components: a phasic component re-
ferred to as skin-conductance response (SCR) and a tonic
component referred to as SCL. Skin-conductance responses
change over the course of seconds and are often used to
measure responses to a novel stimulus, while SCL changes
more gradually and is often associated with tonic states of
sympathetic arousal
21
(such as when a person is stressed).
The ZIPERS, as used in previous research
2,9,10
on the restor-
ative effects of nature, was used to measure affect and aug-
ment physiological measures. Performance on a standardized
mental-arithmetic quiz was used as a measure of cognitive
fatigue. SCL, heart rate, and negative affect scores on the
ZIPERS were hypothesized to significantly decrease, while
positive affect scores on the ZIPERS and mental-arithmetic
performance were hypothesized to increase after exposure to
a nature setting in VR.
Method
Participants
Undergraduate students aged between 17 and 26 from the
University of Waterloo taking an introductory psychology
class were prescreened for trait-immersive tendencies using
the Immersive Tendencies Questionnaire (ITQ) created by
Witmer and Singer.
20
A random sample of 22 students who
scored within one standard deviation of the mean on the ITQ
were asked to participate in the experiment in exchange for
course credit. Participants were recruited and assigned ran-
domly to one of the two conditions (control or nature). There
were four males and six females in the control condition, and
six males and six females in the nature condition. Participants
came to the lab one at a time by scheduled appointments over
the course of 3 weeks.
504 VALTCHANOV ET AL.
Design
A mixed 22 (timecondition) repeated measures design
was used. Measures of restoration (ZIPERS score, SCL, heart
rate, and mental-arithmetic score) were administered before
(time 2) and after (time 3) the VR experience, as shown in
Figure 1. Each participant served as his or her own control on
the repeated measures. The VR experience consisted of par-
ticipants putting on a head-mounted display (HMD) and
observing either a natural setting or abstract paintings, de-
pending on their assigned condition (control vs. nature).
Procedures were identical between conditions, while the VR
tasks each featured a different VR experience. While wearing
the HMD, participants in the nature condition were exposed
to a virtual nature setting, while those in the control condition
were exposed to a slideshow of abstract paintings inside an
empty virtual room featuring only a screen.
Validation of abstract paintings
The abstract paintings used in this study were validated in
a separate pilot study (n¼6) prior to this experiment. In the
pilot study, individuals were shown 30 abstract paintings and
asked to rate them. Ten of the paintings that were consistently
rated by participants as neutral were selected and used in the
control condition of this study.
Measures
Restorative effects were measured in three ways: the ZI-
PERS, heart rate and SCL, and two short mental-arithmetic
quizzes. The ZIPERS
19
was adapted from a similar experi-
ment conducted by Hartig et al.
2
using real natural settings,
and included 12 5-point items that measured the five factors
of fear, anger, positive affect, focus, and sadness, giving in-
dividual scores for each.
19
This scale was found to be an ef-
fective and reliable measure for restoration in previous
research.
2,9,10
Heart rate and SCL were both recorded by a computer
from the start to the end of the experiment. Measurements
were recorded using the PowerLab Data Acquisition System
(ADInstruments, Colorado Springs, CO) and accompanying
LabChart software. SCL was recorded every 10 milliseconds
using two fingertip electrodes placed on the index and mid-
dle fingers of the individual’s non-dominant hand. Heart rate
was recorded every 10 milliseconds using a fingertip sensor
placed on the ring finger of the non-dominant hand. Both SCL
and heart rate have been found to correlate with physiolog-
ical state and stress levels, and have been used in previous
research on restorative effects of natural settings by Ulrich
et al.
10
The mental-arithmetic quizzes featured simple multipli-
cation and division questions. Each quiz contained five
multiplication and five division questions. The quizzes were
designed to be of equal difficulty, both featuring similar types
of problems. One quiz was administered before, and one after
the VR experience (as shown in Figure 1).
Immersive tendencies of participants were also measured
using a standardized ITQ developed by Witmer and Singer.
20
Immersive tendencies are believed to be enduring traits that
are constant over time.
20
These traits are believed to be di-
rectly proportional to how engaged, or ‘‘immersed,’’ an in-
dividual becomes in an attention-demanding task (such as
reading, watching television, or performing a task in VR).
20
The ITQ was used as a control factor, and was administered
within a mass-testing questionnaire prior to recruitment.
Only participants who scored within one standard deviation
of the mean were recruited. This was done to ensure that the
participants in the current experiment did not significantly
vary in immersive tendencies.
Technical information about the virtual-reality setup
The virtual nature setting was constructed using the Elder
Scrolls IV: Oblivion world construction set (Bethesda Soft-
works LLC, Rockville, MD) to handcraft an area equivalent to
1600 m
2
. The area was designed to look like a photo-realistic
forest (see Figure 2). The nature scene was rendered using
the Elder Scrolls IV: Oblivion engine using a GeForce 8800
(nVIDIA, Santa Clara, CA) at 12801024 resolution per eye
with realistic shadows and lighting. The rendered scene was
then piped in stereo to a high-resolution HMD (nVIS, Reston,
VA), capable of displaying a 65-degree field-of-view. The
HMD featured a thick light-blocking cover, which prevented
FIG. 1. Timeline of the events followed in the procedure, shown from left to right. SCL and heart rate were recorded at the
start (time 1), while ZIPERS1 was administered (time 2), and while ZIPERS2 was administered (time 3). The ZIPERS1 was
administered after math quiz 1, in order to measure participants’ affect at their highest level of stress. The ZIPERS2 was
administered right after the VR task, but before math quiz 2 (because the second math quiz would have contaminated the
effect of the VR immersion by stressing participants). The ZIPERS was administered in such a way that it ‘‘sandwiched’’ the
VR manipulation (as shown), giving a clear measure of ‘‘before’’ and ‘‘after’’ VR immersion.
RESTORATIVE EFFECTS 505
participants from seeing the real-world environment around
them, allowing them to focus on the virtual scene presented.
The viewpoint was calibrated to work with an InertiaCube
(InterSense Inc., Billerica, MA) tracking device attached to the
HMD. This setup allowed for the viewpoint to be updated in
real time, such that it mimicked physical head movements.
For example, if the participant rotated his or her head to the
left, the viewpoint in the virtual environment would update
accordingly and look toward the left.
A large rumble platform was also incorporated into the
VR setup, in order to provide somatosensory stimulation dur-
ing experiences within the virtual environment. The rumble
platform was 2.5 meters in diameter, and was configured to
vibrate whenever a ‘‘step’’ was taken within the virtual en-
vironment, as well as whenever there was impact between
the virtual body and an object (i.e., when falling off an ele-
vation or when bumping into a tree).
Self-locomotion within the environment was done through
use of a wireless mouse. The left mouse button was config-
ured to move the virtual body forward, while the right mouse
button was configured to move the virtual body backward.
Turning within the virtual environment required the partic-
ipant to turn their physical body. The direction of movement
within the virtual world was the same as the direction of the
physical body of the participants in the real world.
Procedure
Participants were randomly assigned to one of the two
conditions (control or nature). An identical procedure (as
shown in Figure 1) was followed for both conditions with the
exception of the manipulation itself. Participants were gree-
ted in a centralized meeting area and escorted individually to
the lab by the experimenter. They were then asked to read
and sign a consent form that described all procedures in the
experiment. Participants were given a cover story stating that
the purpose of the experiment was to examine the effects of
VR on math-test performance.
After the experiment was explained to the participants,
heart-rate and SCL monitoring electrodes were placed on
their non-dominant hand. A 2-minute baseline was recorded
(time 1 in Figure 1). They were then asked to put on stereo
headphones, and asked to describe (in writing) a stressful
event that had occurred to them in the last 6 months, while
loud urban noise (cars honking, jackhammers working, etc)
was played through the headphones. Participants were given
10 minutes to complete this stress-induction task (as indicated
in Figure 1). Upon completion of the task, participants were
given one of two short mental-arithmetic quizzes and given
5 minutes to complete it (‘‘math quiz 1’’ in Figure 1—head-
phones were removed before this was done). Participants
were instructed to complete as many of the math problems as
they could in the allotted time. The purpose of this was to
increase stress levels beyond those normally experienced, as
well as to cause a significant amount of cognitive fatigue. This
method of stress induction was inspired by the Markus &
Peters Arithmetic (MPA) Test,
18
which features a stressful
mental-arithmetic task. During the MPA Test, participants
are required to complete difficult mental-arithmetic problems
while listening to industrial noise. The MPA Test has been
used by de Kort,
16
and has been found to increase heart rate
and SCL, and to induce a negative mood.
16,18
After the 5 minutes had passed, participants were asked to
complete the ZIPERS (ZIPERS 1 at time 2 in Figure 1). While
they were answering the ZIPERS, they were asked to place
their non-dominant hand (which was attached to the elec-
trodes) on the table and not to move it. This allowed physio-
logical data to be collected without the movement of artifacts
prior to entering VR. Participants were then asked to put on
a HMD and stand on the rumble platform. The rumble plat-
form was only turned on in the nature condition. The ma-
nipulation was then applied, with the experience for each
participant while wearing the HMD depending on the con-
dition to which they were randomly assigned.
In the control condition, participants watched a slideshow
of abstract paintings for 10 minutes. There were 10 paintings
in the slideshow, and each was displayed for exactly 1 min-
ute. Paintings featured large proportions of common colours
found in nature, such as green (normally seen on trees, grass,
and plants) and blue (normally seen in the sky and water), as
well as very small proportions of other colours (red, orange,
and yellow) that are normally seen in flowers and fruits.
Participants had no control over the slideshow or the order of
the paintings. Paintings were presented in a predefined,
randomly generated order. The slideshow included a large
virtual screen (similar to those used for real-world slide-
shows) that displayed each painting for a set amount of time
within a dark virtual room with the lights off. See Figure 3 for
a sample of the paintings used.
In the nature condition, participants freely explored a
nature setting, which was rendered in real time. The setting
consisted of a photo-realistic forest with various shrubs,
flowers, trees, and small bodies of water (including water
streams and small ponds; see Figure 2). A variety of large and
small rocks and terrain elevations were also present. The
scene was rendered in high resolution (12801024 pixels per
eye) with realistic lighting, ambient natural sounds playing
in the background, and autonomous movement (plants
and trees moving with the wind, and ripples in the water).
Somatosensory feedback was also provided using a large
rumble pad on which participants stood. The rumble pad
shook slightly every time they took a ‘‘step’’ in VR. A ‘‘Forest
Breeze’’ Wick air freshener was also used, in an attempt to
provide participants with olfactory cues associated with be-
ing in a forest. Participants were given complete control over
the viewpoint by use of head-tracking technology, which
updated the viewpoint according to their physical head
movements.
FIG. 2. Screen capture of the three-dimensional photo-
realistic forest that was constructed for the high-presence
restorative task. The entire forest (1600 m
2
) could be explored
freely.
506 VALTCHANOV ET AL.
Participants wore the HMD and observed the virtual set-
tings for 10 minutes in both conditions (referred to as VR
immersion in Figure 1). After this task was completed, the
HMD was removed, and participants were asked to sit at a
desk and complete a second ZIPERS scale (ZIPERS2 at time 3
in Figure 1). During this time, participants were again asked
to place their non-dominant hand, which was connected to
the SCL and heart-rate monitors, on the table and to refrain
from moving it so that physiological data could be collected.
After this, participants were asked to complete the second
math quiz within 5 minutes (‘‘math quiz 2’’ in Figure 1), and
were then debriefed. A timeline summary of the procedure
can be seen in Figure 1.
Results
Statistical analysis of immersive tendencies
An empirical check was implemented to ensure that par-
ticipants in the two conditions did not differ in immersive
tendencies, since differences could have potentially impacted
their immersion within the virtual environments. The ITQ
was administered to participants prior to their participa-
tion in the study. A one-way ANOVA revealed no signifi-
cant differences between the two conditions, F(1, 20) ¼0.54,
Z
p2
¼0.027, p¼0.468, indicating that immersive tendencies of
participants were constant (as intended) between groups.
Statistical analysis of physiological measures
The physiological state of participants was measured
using SCL and IBI. Measurements were taken continuously
throughout the duration of the experiment, with the baseline
established at the start of the experiment before participants
performed any tasks. The measurements were analyzed at
three critical times of interest: at the start of the experimental
session (time 1); just before the manipulation of immersion
into VR, and just after the stress-induction manipulation
(time 2); and just after the VR exposure (time 3). Measure-
ments of both SCL and IBI were averaged over a 2-minute
interval surrounding the time of interest (e.g., time 2 and
time 3). Figure 1 provides a chronological diagram of when
measures were administered. For statistical analysis, each
participant served as their own control. A baseline for HR
and SCL was established for each participant at time 1, and
the monitoring equipment was zeroed to this baseline. Since
each participant had their own unique SCL baseline, SCL
measurements for individual participants were converted
into z-scores to allow for between-subjects comparisons.
In order to confirm that the stress-induction manipulation
was effective, the data was collapsed across groups (since all
groups experienced the same stress-induction manipulation),
and two repeated-measures analyses of variance (ANOVAs)
were used to analyze the change in SCL and IBI from time 1
(the start) to time 2 (post-stressor). SCL was found to in-
crease significantly from time 1 to time 2, F(1, 21) ¼31.57,
MSE ¼0.514, Z
p2
¼0.601, p0.001. A trend decrease was
found in IBI (indicating a higher heart rate) from time 1 to
time 2, F(1, 21) ¼3.152, MSE ¼0.002, Z
p2
¼0.131, p¼0.09. To
verify that there were no differences in IBI and SCL at time 2
(post-stressor), a one-way ANOVA examining IBI and SCL
between conditions was performed. No differences between
conditions were found in SCL, F(1, 21) ¼0.575, n.s., and
no differences in IBI were found, F(1, 21) ¼0.391, n.s.,
confirming that SCL and IBI were not different between
groups before immersion into VR (as intended). These find-
ings correspond with the notion that SCL increases and IBI
FIG. 3. Three organic art paintings that were presented during the slideshow of the control condition.
RESTORATIVE EFFECTS 507
decreases as personal levels of stress and cognitive arousal
increase.
16,21
Two repeated-measures mixed ANOVAs were used to
examine the effect of the VR manipulation (abstract art sli-
deshow vs. nature environment) on IBI and SCL from time 2
(post-stress) to time 3 (after VR). In the first ANOVA, IBI
was used as the within-subjects dependent variable, and the
condition (control vs. nature) was used as the between-
subjects variable. The second ANOVA used the same inde-
pendent variable as the first ANOVA, but used SCL as the
within-subjects dependent variable. The first repeated-
measures mixed ANOVA revealed no significant interaction
between the type of manipulation (nature exposure vs. slide-
show of abstract art) and change in IBI, F(1, 20) ¼0.621,
p¼0.440, n.s. This indicated that the change in IBI from time 2
to time 3 did not differ between conditions. These results are
shown in Figure 4. The second repeated-measures mixed
ANOVA revealed a significant interaction between the type
of VR manipulation and SCL, F(1, 20) ¼15.072, MSE ¼0.466,
Z
p2
¼0.430, p¼0.001, suggesting that the VR manipulations
had different effects on SCL. This warranted further investi-
gation of simple effects.
To test for simple effects, the data was split with respect
to condition (nature vs. control), and repeated-measures
ANOVAs were used to examine the effect of VR on SCL. A
significant decrease in SCL after exposure to the nature vir-
tual environment was found, F(1, 11) ¼36.082, MSE ¼0.514,
Z
p2
¼0.766, p<0.001. However, the abstract-art slideshow
had no effect on SCL, F(1, 9) ¼0.293, p¼0.601, n.s. These
results are shown in Figure 5.
Statistical analysis of self-report measures
ZIPERS measure. To augment physiological measures,
self-report measures were also collected. The ZIPERS was
administered at time 2 and time 3, before and after the VR
manipulation (see Figure 1). The purpose of this was to ex-
amine the effects of the VR manipulation on the five factors
(fear, anger, focus, positive affect, and sadness) measured by
the ZIPERS. For the purpose of analysis, sadness, fear, and
anger were grouped into one category, which we defined as
‘‘negative affect,’’ leaving us with three factors of interest
(positive affect, negative affect, and focus). The items grouped
into ‘‘positive affect’’ were found to have a Cronbach’s alpha
of 0.825, while items grouped into ‘‘negative affect’’ were
found to have a Cronbach’s alpha of 0.740. A series of
repeated-measures mixed ANOVAs with VR manipulation
(nature vs. slideshow) as the between-subjects independent
variable, and VR immersion (time 2 to time 3; immersion into
VR) as the within-subjects independent variable were used to
examine scores for the three relevant factors (positive affect,
negative affect, and focus).
The VR manipulation by positive-affect interaction was
found to be trending significance, F(1, 20) ¼3.925, MSE ¼
0.463, Z
p2
¼0.164, p¼0.061, suggesting that there was likely
an effect of the VR manipulations on change in positive affect
from time 2 to time 3. This warranted further exploration
of simple effects to disambiguate the direction of the trend-
ing effects. Simple effects were tested by splitting the data
with respect to condition (nature vs. control), and perform-
ing repeated-measures ANOVAs on positive affect over time
(time 2 to time 3). In the nature condition, positive affect was
found to increase significantly, F(1, 11) ¼8.758, MSE ¼0.743
Z
p2
¼0.443, p<0.05. Meanwhile, in the control condition,
there was no significant change in positive affect, F(1, 9) ¼
2.089, n.s. Unlike positive affect, the VR manipulation by
negative-affect interaction was not significant, F(1, 20) ¼
0.365, p¼0.553, n.s., indicating that the VR manipulations did
not promote a different change in negative affect from time 2
to time 3. Also, no VR manipulation by focus interaction was
found, F(1, 20) ¼0.430, p¼0.519, n.s. The results for positive
and negative affect are shown in Figure 6.
FIG. 4. Graph of change in IBI in milliseconds at each critical time: at the start of the experiment (time 1) where the baseline
was established, after the stress-induction task (time 2), and after the VR manipulation (time 3), for the two conditions. Here,
it can be seen how IBI trended decrease from time 1 (start) to time 2 (post stressor) as a result of the stressor task within each
restorative task.
508 VALTCHANOV ET AL.
Math-quiz measure. Two mental-arithmetic quizzes were
also administered before and after immersion into VR, in
order to test potential increase in focus and recovery from
cognitive fatigue. The quizzes were counterbalanced be-
tween participants, and consisted of five multiplication and
five division questions. A repeated-measures ANOVA was
used to test for differences in math score with respect to time.
There were no significant differences in math scores before
(M¼75%, SD ¼23%) and after (M¼79%, SD ¼18%) im-
mersion into VR, F(1, 21) ¼2.8, MSE ¼5.114, p¼0.109, n.s.
Simple-effects analyses were done using independent sample
t-tests to check for expected gender effects, and to confirm
that the two math quizzes did not differ in difficulty. In-
dependent sample ttests revealed that males performed
significantly better than females before, t(31) ¼2.98, p<0.01,
and after, t(31) ¼2.11, p<0.05, immersion into VR. Since
groups contained roughly equal numbers of males and fe-
males (45% and 55% respectively), gender differences were
discounted as a source of significant variation between
groups. No significant differences between the two math
quizzes were found at time 2 (prior to immersion in VR),
indicating that the math quizzes were equally difficult,
t(20) ¼0.468, SE ¼0.996, p¼0.64, n.s.
Discussion
As seen in Figures 5, and 6, SCL decreased significantly
and positive affect increased significantly after exposure to
computer-generated nature, supporting our first hypothesis that
computer-generated nature can promote restorative effects.
In the current experiment, the effects of exposure to com-
puter-generated nature in VR appear to be consistent with
the expanded model of restorative effects elicited by both
real nature settings
2,5
and surrogate nature such as photo-
graphs
9,12
and color videos.
10,15,16
There are several implica-
tions of these results.
First, the consistency with previous research using surro-
gate nature such as that by Ulrich,
10
van den Berg et al.,
15
and
de Kort et al.
16
provides further support for the notion that
restorative effects (such as improved affect and decreased
physiological stress) can be elicited by surrogate nature. In
the case of the current study, an increase in positive affect and
decrease in SCL demonstrate that computer-generated na-
ture, which is not a direct replication (such as a photograph or
video) but rather an artistic interpretation of what trees, water,
sky, light and other nature elements look and sound like, can
promote restorative effects. These results are also somewhat
FIG. 5. Graph of z-scores of the SCL measure at each time: at the start of the experiment (time 1) where the baseline was
established, after the stress-induction task (time 2), and after the VR manipulation (time 3), for each of the restorative tasks.
Here, it can be seen how SCL increased significantly from time 1 (start) to time 2 (post-stressor) as a result of the stress-
induction task in both conditions. Differences at time 2 (post-stressor) were not significant. By comparing time 2 (post stressor)
and time 3 (after VR), it can be seen that SCL did not decrease in the control condition while it decreased significantly in the
nature condition.
FIG. 6. Graph of the mean ZIPERS scores for positive affect
and negative affect at time 2 (post stressor) and time 3 (after
VR). Positive affect increased significantly in the nature
condition, while it did not change significantly in the control
condition. Here, the floor effect in the negative affect scores
can be also seen.
RESTORATIVE EFFECTS 509
consistent with other studies using computer-generated na-
ture,
22,23
which have demonstrated increased self-reports of
feeling relaxed after a relaxation session incorporating a vir-
tual island. However, due to large differences in both the
procedure (using a relaxation session without a prior stress
induction) and the nature stimulus (beaches on an island vs.
a forest), further generalizations cannot be made.
In the current study, participants’ knowledge about the
synthetic properties of the nature environment did not ap-
pear to negate the beneficial effects of taking a walk inside
a virtual forest. This suggests that our predisposition to fa-
voring nature may not necessarily be confined to what is
‘‘real’’ and=or tangible, but rather that it may include syn-
thetic elements (such as three-dimensional re-creations like
those in VR, and photo=video-graphic replications). This
idea, that important restorative elements of nature can be
captured using media, is supported by previous research
using videos of nature settings to promote restoration.
10,15,16
It must be noted that the experimental procedure followed
in the current study does not allow for the differentiation
between attribution of the results to virtual nature settings,
and virtual settings in general. Although the results are
consistent with predicted restorative effects of surrogate na-
ture, one cannot discount the possibility that VR has restor-
ative properties in general, since the control condition did not
feature a full VR setup that allowed participants to explore
and interact with an environment. In the present study, we
did not use an interactive virtual environment for the control
condition because creating such an environment posed many
challenges that could not be rectified. Urban settings have
been used as the standard control in restorative-effects liter-
ature.
10,11,14
However, the continued use of urban settings as
a control has created a problem: It is still unclear whether
nature has restorative properties, or whether urban settings
simply induce stress. Thus the current study did not use an
interactive urban setting in order to avoid this ambiguity. In
not doing so, we were left with the alternative of attempting
to create an abstract virtual environment that did not feature
a nature or urban setting. However, in creating such an en-
vironment, one must consider the repercussions from im-
mersing individuals in unrealistic virtual environments, and
how individuals would respond to a virtual environment that
does not possess the qualitative properties of the real world.
After considering the potential for serious confounds that
such an environment would introduce, we opted to use a
‘‘safer’’ control that did not feature interactive VR but was still
realistic (i.e., looking at an abstract painting in VR is more
realistic than exploring a virtual world that looks like an
abstract painting). Unfortunately, using a control condition
that did not feature an interactive VR experience introduced
the confound suggesting that the interactive portion of the
nature condition may have contributed to the size of the ef-
fects. However, previous research by de Kort et al.
16
dem-
onstrating the restorative properties of nature in the absence
of interactive environments lead us to believe that if the in-
teractive properties of VR could not be solely responsible for
the restorative properties of the nature condition in the
present study.
When comparing the physiological responses to the control
and nature conditions, there are some oddities. As shown in
Figure 4, IBI increased (i.e., heart rate decreased) at approxi-
mately the same rate in both conditions from time 2 to time 3,
suggesting that the abstract-art slideshow and nature envi-
ronment had similar effects on IBI. The lack of interaction
between the type of virtual environment and the change in
IBI was contrary to a priori predictions stating that IBI would
increase more from exposure to nature when compared to a
control group. There are several explanations that could ac-
count for this. It is possible that the abstract-art slideshow
provided an opportunity for participants to relax, resulting in
a similar increase in IBI as in the nature condition. It is also
possible that IBI increased as it returned to baseline over time
as the effects of the stress-induction faded. However, this is a
weaker explanation that does not necessarily explain the
pattern shown in Figure 4, where IBI appears to be above
baseline at time 3. Lastly, it is possible that the VR experience
itself prompted changes in IBI. However, such a case seems
unlikely, as we know of no previous evidence to suggest that
all VR environments have restorative properties. Future work
investigating the potential benefits of immersion into VR is
needed to verify this. Also, future work comparing VR re-
storative effects to effects of other restoration methods (clos-
ing eyes and relaxing, listening to nature sounds, looking at
photos and videos, and going out for a real-world nature
walk) needs to be done to better gauge the validity of resto-
ration achieved through VR.
The lack of difference in math scores before and after im-
mersion was also contrary to what was expected. There are
several possible explanations for these results. The first sug-
gests that math quizzes were a poor choice of measure for
focus and cognitive fatigue, since math ability may not be
affected by such factors. In retrospect, a reading compre-
hension task, such as that used by Hartig et al.,
2
may have
been a better measure. The second explanation hints at the
possibility that the stress-induction writing task was not
cognitively taxing enough to result in cognitive fatigue and
loss of focus, leaving participants at near-baseline cognitive
ability with little room for recovery. Another possible expla-
nation is that immersion into VR does not encourage recovery
of cognitive ability. Lastly, it is also possible that the math
quizzes were not difficult or long enough (as shown by the
high means) to adequately differentiate between emotional
and cognitive states. Unfortunately, due to the ambiguity of
the math results, it is impossible to determine which of these
reasons may be the source of the unexpected results.
Finally, the lack of differences in change of negative affect
from time 2 to time 3 was also contrary to a priori predictions.
However, by looking at Figure 6, it can be seen that there was
a floor effect in the ZIPERs measure. Participants rated neg-
ative-affect factors on a scale of 1 (none) to 5 (very high), and
scores at time 2 were already close to 1 (i.e., there was no
negative affect). This left little room for scores to improve.
Further research exploring the possibility that virtual nature
environments can also reduce negative affect is required. A
more rigorous stress-induction task that induces strong neg-
ative affeect, such as the MPA Test,
18
might help create a level
of negative affect that can disambiguate differences between
a control virtual environment and a virtual environment
featuring a nature setting.
Implications for future research
The current study provides a new tool for finding ways to
understand ‘‘how’’ and ‘‘why’’ exposure to nature promotes
510 VALTCHANOV ET AL.
restorative effects. By showing results consistent with previ-
ous work,
10,16
and demonstrating that restorative effects can
be elicited using computer-generated nature and VR, the
current study suggests that VR can be used to understand
better how and why restorative environments function. With
the help of VR technology, restorative environments can be
constructed, as in the current study, and then systematically
changed in stages to explore how these changes affect resto-
ration. Similarly, non-restorative environments can be con-
structed and changed in stages until they become restorative.
Through the use of such methods with VR, future research
can begin to examine specific aspects of what creates a re-
storative environment and what creates a non-restorative or
stressful environment, while maintaining a high level of ex-
perimental control. Virtual city downtown cores can be filled
with trees and plants at the push of a button within a con-
trolled lab setting, while also maintaining complete control
over other aspects of the environment (such as traffic,
streetlights, advertisements, etc). Meanwhile, virtual restor-
ative forest or park environments can be constructed and then
systematically changed to included man-made features such
as paved paths, water fountains, lampposts, and other urban
features. The power over manipulating the environment in a
controlled manner (without the need for construction crews
to dig and modify a landscape), combined with the ease and
time efficiency, makes VR a very powerful research tool. With
such a powerful tool in the hands of researchers, it may be
possible to understand further why nature is important, and
what can be learned from natural environments.
With modern advancements in mainstream VR technol-
ogy, VR setups such as the one used in this experiment
have become available to consumers. Currently, consumers
may purchase a HMD with head-tracking that can attach to
their regular desktop computer for under $300 USD. With
the emergence of such technologies that allow regular in-
dividuals to experience immersive worlds, it is vital for
research to explore further the beneficial uses of VR. The
rising popularity of three-dimensional glasses and HMDs
in the entertainment industry gives rise to an intriguing
possibility of improving psychological health through re-
storative environments. This possibility further stresses the
importance of the findings presented in the current study,
and of future research exploring restorative environments
using VR.
In conclusion, the results of the current study demonstrate
that computer-generated nature in VR can promote restor-
ative effects, and thus provides the field of restorative re-
search with a new empirical tool that can help researchers
dissect the makings of both restorative and non-restorative
environments with full customizability and a high degree of
experimental control that would not otherwise be possible.
With such a tool at their disposal, researchers may not only
gain a deeper understanding into what constitutes a restor-
ative environment, but also develop methods and environ-
ments that may have therapeutic applications in scenarios
where real nature cannot be directly accessed or incorporated
into interior settings.
Acknowledgments
The present research was supported by funding awarded
to Colin Ellard by the Natural Sciences and Engineering Re-
search Council of Canada (NSERC) and the Social Sciences
and Humanities Research Council of Canada (SSHRC).
Disclosure Statement
No competing financial interests exist.
References
1. Gullone E. Biophilia hypothesis in the 21st century: In-
creasing mental health or increasing pathology? Journal of
Happiness Studies 2000; 1:293–321.
2. Hartig T, Mang M, Evans GW. Restorative effects of natural
environment experiences. Environment & Behavior 1991; 23:
3–26.
3. Kaplan R. Some psychological benefits of an outdoor chal-
lenge program. Environment & Behavior 1974; 6:101–16.
4. Kaplan R. Effects of grouping and response characteristics
of instructional objectives when learning from prose. Journal
of Educational Psychology 1976; 68:424–30.
5. Berman MG, Jonides J, Kaplan S. The cognitive benefits
of interacting with nature. Psychological Science 2008; 19:
1207–12.
6. Kaplan S. The restorative benefits of nature: Toward an in-
tegrative framework. Journal of Environmental Psychology
1995; 15:169–82.
7. Kaplan S. Meditation, restoration, and the management of
mental fatigue. Environment & Behavior 2001; 33:480–506.
8. Berto R, Massaccesi S, Pasini M. Do eye movements mea-
sured across high and low fascination photographs differ?
Addressing Kaplan’s fascination hypothesis. Journal of En-
vironmental Psychology 2008; 28:185–91.
9. Ulrich RS. Natural versus urban scenes: Some psycho-
physiological effects. Environment & Behavior 1981; 13:
523–56.
10. Ulrich RS, Simons RF, Losito BD, et al. Stress recovery
during exposure to natural and urban environments. Journal
of Environmental Psychology 1991; 11:201–30.
11. Ulrich RS, Simons RF, Miles MA. Effects of environmental
simulations and television on blood donor stress. Journal of
Architectural & Planning Research 2003; 20:38–47.
12. Kweon B, Ulrich RS, Walker VD, et al. Anger and stress: The
role of landscape posters in an office setting. Environment &
Behavior 2008; 40:355–81.
13. Leather P, Pyrgas M, Beale D, et al. Windows in the work-
place: Sunlight, view, and occupational stress. Environment
& Behavior 1998; 30:739–62.
14. Weinstein N, Przybylski A, Ryan R. Can nature make us
more caring? Effects of immersion in nature on intrinsic
aspirations and generosity. Personality & Social Psychology
Bulletin 2009; 35:1315–29.
15. van den Berg AE, Koole SL, van der Wulp NY. Environment
preference and restoration: (How) are they related? Journal
of Environmental Psychology 2003; 23:135–46.
16. de Kort YAW, Meijnders AL, Sponselee AAG, et al. What’s
wrong with virtual trees? Restoring from stress in a medi-
ated environment. Journal of Environmental Psychology
2006; 26:309–20.
17. de Kort YAW, Ijsselsteijn WA. Reality check: The role of
realism in stress reduction using media technology. Cyber-
Psychology & Behavior 2006; 9:230–33.
18. Peters ML, Godaert GL, Ballieux RE, et al. Cardiovascular and
endocrine responses to experimental stress: Effects of mental
effort and controllability. Psychoneuro-endocrinology 1998;
23:1–17.
RESTORATIVE EFFECTS 511
19. Zuckerman M. Development of a situation-specific trait–
state test for prediction and measurement of affective re-
sponses. Journal of Consulting & Clinical Psychology 1977;
45:513–23.
20. Witmer B, Singer M. Measuring presence in virtual environ-
ments: A presence questionnaire. Presence 1998; 7:225–40.
21. Dawson ME, Schell AM, Filion DL. (2007) The electrodermal
system. In: Cacioppo JT, Tassinary LG, Bernston GG, eds.
Handbook of psychophysiology. 3rd ed. New York: Cambridge
University Press, pp. 159–81.
22. Freeman J, Lessiter J, Keogh E, et al. Relaxation island:
Virtual, and really relaxing. Paper presented at: 7th Inter-
national Workshop on Presence; October 13–15, 2004; Uni-
versitat Politecnica de Valencia, Spain.
23. Villani D, Riva F, Riva G. New technologies for relaxation:
The role of presence. International Journal of Stress Man-
agement 2007; 14:260–74.
Address correspondence to:
Deltcho Valtchanov
Department of Psychology
University of Waterloo
Waterloo, ON
Canada
N2L 3G1
E-mail: deltcho@gmail.com
512 VALTCHANOV ET AL.