See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260511298

Physiological and affective responses to immersion in virtual reality: Effects

of nature and urban settings

Article · January 2010

CITATIONS READS

37 1,301

1 author:

Deltcho Valtchanov
University of Waterloo
17 PUBLICATIONS 834 CITATIONS

SEE PROFILE

All content following this page was uploaded by Deltcho Valtchanov on 05 March 2014.

The user has requested enhancement of the downloaded file.

Journal of CyberTherapy & Rehabilitation 359
W i n t e r 2 0 1 0 , Vo l u m e 3 , I s s u e 4
© Vi r t u a l R e a l i t y M e d i c a l I n s t i t u t e

physiologiCal and affeCTiVe responses

To immersion in VirTual realiTy:
effeCTs of naTure and urban seTTings
Deltcho Valtchanov1 and Colin Ellard1

With the rapid advancements in technology, researchers are seeking new ways to incorporate modern
high-tech solutions such as Virtual Reality (VR) into treatment paradigms for stress. The current ex-
periment explores the beneficial effects of immersing an individual into VR after a stressful encounter.
The potential restorative effects of three unique immersive VR environments were examined by induc-
ing stress and negative affect in 69 participants, and then randomly assigning them to freely explore
one of three environments (a virtual nature setting, a virtual urban cityscape, or a neutral environment
composed of solid geometric shapes) for 10 min. Participants who explored the nature environment
were found to have significantly improved affect (as measured by a standardized questionnaire), and
significantly lower stress levels (as measured by self-report and skin conductance levels) compared to
those who explored the urban and geometric environments. The results suggest that virtual nature has
restorative properties similar to real nature, and that simply immersing participants into a virtual nature
setting can reduce stress. These results also suggest that the content of the VR experience (i.e., whether
it contains nature) is important in promoting restoration, and that in the absence of nature, stress levels
remain unchanged.

Keywords: Biophilia, Restorative Effects, Stress Management, Virtual Reality, Nature

inTroduCTion Network Task) improved after viewing pictures of nature

Over the last 30 years, researchers have provided empiri- when compared to individuals who viewed pictures of
cal evidence suggesting that surrounding oneself with na- urban areas (Berman et al., 2008) which suggests that sim-
ture can have restorative effects on emotional, ply viewing nature pictures can have restorative effects.
physiological and cognitive states. The restorative effects Other studies by Ulrich et al. (Ulrich, 1981; Ulrich, Si-
of exposure to nature have been classified as a reduction mons, & Losito, 1991) have also shown that viewing pho-
in cognitive fatigue, decreases in both physiological and tographs and videos of nature scenes can promote
cognitive stress, a decrease in negative affect, and an in- significant reductions in physiological stress (shown by
crease in positive affect (Gullone, 2000; Hartig, Mang & reductions in skin conductance level) and improvements
Evans, 1991; Berman, Jonides & Kaplan, 2008). For ex- in emotional states of individuals (shown through self-re-
ample, recent research by Berman, Jonides and Kaplan port on the Zuckerman Inventory of Personal Reactions).
(2008) comparing the effects of interacting with natural
versus urban environments has found that individuals per- Kaplan (1995, 2001) delineates how and why nature is
formed better on a working memory task after taking a restorative in his Attention Restoration Theory (ART),
walk in a local park versus taking a walk in a local down- which states that when a person is immersed and interact-
town area. Furthermore, Berman et al. found that individ- ing with a surrounding environment that contains fasci-
uals’ executive attention (as measured by the Attention nating stimuli, the stimuli modestly capture attention in an

Corresponding Author:
Deltcho Valtchanov, Research Laboratory for Immersive Virtual Environments (RELIVE), Department of Psychology, University of Waterloo, Wa-
terloo, Ontario, Canada, N2L 3G1, E-mail: [email protected]

1
University of Waterloo, Waterloo, Ontario, Canada, N2L 3G1
360 Physiological and Affective Responses to VR

involuntary fashion. ART further states that while this theory. These authors have shown that exposure to surro-
form of involuntary attention is active, internal mecha- gate nature (such as videos and photos of real nature) pro-
nisms responsible for directed attention are allowed to re- moted a decrease in heart rate and skin conductance
cover (Kaplan, 1995; Kaplan, 2001). ART attempts to (Ulrich et al., 1991), and even blood pressure (Ulrich, Si-
explain the differences between exposure to nature versus mons & Miles, 2003) while also improving self-reported
environments by contrasting the types of stimuli present affect.
in both types of environments – nature environments con-
tain many stimuli that modestly capture attention (such as Much of the research on the restorative effects of nature
grass and leaves swaying with the wind), which occupy has focused on identifying and understanding how the
attention but can also be disengaged from easily, allowing restorative properties of nature may be utilized to improve
attention to drift from stimulus to stimulus. Conversely, overall quality of life and well-being, rather than exploring
urban environments contain bright lights, neon signs, loud the theories proposed by Kaplan (1995; 2001) and Ulrich
vehicles and construction sounds that dramatically capture et al. (1981; 1991). Some applications include making
attention. Thus, directed attention mechanisms are taxed cityscapes more pleasant and appealing to individuals by
in order to disengage from the stimuli (e.g. ignoring neon adding trees (Sheets & Manzer, 1991) while others include
advertisements), causing the environments to be less improving workplace satisfaction by providing views of
restorative. (Kaplan, 1995; Kaplan, 2001). Recent research natural scenery within office spaces (be it through the use
by Berto Massacesi, and Pasini (2008) examining eye of windows or posters) (Leather, Pyrgas, Beale, &
movements of participants that viewed urban and nature Lawrence, 1998; Kweon, Ulrich, Walker, & Tassinary,
images provides some support for ART. Berto et al. (2008) 2008). With the emergence of new technologies such as
found that when participants viewed images of nature immersive Virtual Reality (VR) and 3-D video, researchers
scenes, they had fewer long fixations and visually ex- are investigating new ways to use these technologies to
plored more of the images compared to participants that help improve the quality of life both within the home and
viewed images of urban settings. This provides evidence clinical settings (Villani & Riva, 2008; Valtchanov et al.,
that nature scenes prompt involuntary attentional drifts 2010). Part of this line of research has focused on exploring
from stimulus to stimulus as suggested by ART. the restorative effects of exposure to nature using immer-
sive VR. Valtchanov, Barton and Ellard (2010) believe that
A second theory that stands in contrast to Kaplan’s ART the restorative properties of nature may be captured using
was proposed by Ulrich et al. (1981; 1991). This theory computer-generated nature (i.e., the artistic interpretation
states that an individual’s initial response to immersion of nature versus photographs or videos of real nature)
within an environment is affective instead of cognitive. which can then be used within VR to create restorative ex-
Ulrich et al. proposed that the structural properties of an periences. This notion is supported by research done by
environment (such as complexity and focality) prompt an Villani and Riva (2008) which has found that immersive
automatic affective response within the individual. Ulrich 3-D video (which coincidentally contained computer-gen-
et al.’s theory proposes that stimuli seen outside of nature erated nature) can aid recovery in a stress management pro-
are more threatening and thus more physiologically arous- tocol. Furthermore, research by Villani et al. (Villani, Riva,
ing. In contrast to ART, where replenishment of directed & Riva, 2007; Villani, Luchetta, Preziosa, & Riva 2009)
attention mechanisms is believed to be the source of and Freeman, Lessiter, Keogh, Bond & Chapman (2004)
restoration, Ulrich et al.’s theory proposes that the initial that suggests VR may be used to aid relaxation also used
automatic affective response to an environment shapes the computer-generated nature environments. Lastly,
cognitive events that follow (Valtchanov, Barton, & Ellard, Valtchanov et al. (2010) investigated whether computer-
2010). If the affective response is positive (as when one generated virtual nature had restorative effects, and found
is exposed to nature), the cognitive and physiological that exploring a virtual forest promoted reductions in phys-
events that follow are also positive – negative emotions iological stress, and an improvement in emotional state
and thoughts are suppressed, resulting in higher levels of when compared to viewing abstract paintings. While this
positive affect and ability to sustain attention, and reduced research supports the idea that computer-generated nature
levels of negative affect and physiological stress may possess some of the properties of real nature to pro-
(Valtchanov et al., 2010). Experiments by Ulrich et al. duce restoration, it does not definitively answer the ques-
have focused on measuring the affective and physiological tion of whether it is the nature scenery or the immersive
states of participants in an attempt to find support for this VR experience that is responsible for the observed restora-

JCR
Valtchanov and Ellard 361

tive effects. Such is the case partially due to the lack of con- as proposed by Kaplan (1995; 2001) and the “affective re-
trolled comparison groups. As acknowledged by sponse” theory proposed by Ulrich et al., (1981; 1991)
Valtchanov et al. (2010) it is currently unclear whether the since both theories suggest contradictory outcomes when
restoration produced by virtual nature shares a mechanism participants are exposed to such an “abstract” environ-
with restoration produced by real nature, or whether VR in ment. Kaplan’s ART would suggest that such an environ-
general may provide an “escape” from one’s current mind- ment would be restorative since it lacks elements that
set and situation, resulting in psychological distance from demand attentional resources (such as advertisements,
stressors and thus restoration. loud noises, etc), while still containing a new and fasci-
nating experience (i.e., VR) to modestly capture attention,
The goal of the present study was to build on the previous allowing directed attention to disengage and recover.
research by Valtchanov et al. (2010). Using improved Meanwhile, according to Ulrich et al.’s theory, since the
methodology and measures, this study aspired to test the environment is similar to an urban setting in its geometric
effects of a variety of virtual environments on physiolog- focus, and contains visual patterns that are not found in
ical stress, ability to sustain attention, and affect. One en- nature, it should be viewed as threatening and promote
vironment was a virtual nature environment, another was stress and deterioration of affect. Lastly, the current study
a virtual urban environment modelled after the Shibuya also aimed to examine whether virtual nature could pro-
area in Tokyo Japan, and the last was a geometric virtual mote an improvement in ability to focus, similar to that
environment featuring only regular geometric shapes reported by Berman et al. (2008) when participants were
(such as cubes, cylinders, spheres, etc.). Similar measures immersed in real nature.
to Valtchanov et al. (2010) were used in order to examine
whether the previously reported restorative effects of vir- We predicted that there would be a similar pattern of re-
tual nature settings could be reproduced using a more sults to that observed in research using real nature, such
complicated and realistic nature environment conforming that the virtual nature environment would promote
to the current computing power. This was also done to stronger improvements in physiological, affective, and
compare physiological and affective reactions of partici- cognitive states than the urban and geometric environ-
pants to three distinctly different virtual environments to ments as shown by a decrease in skin conductance level,
clarify whether VR provides an “escape” which may cause reduced heart rate, higher positive affect and lower nega-
restoration, or whether it is virtual nature that has been re- tive affect scores on the Zuckerman Inventory of Personal
sponsible for the restoration observed in previous literature Reactions (ZIPERS), and fewer errors on the Sustained
(Freeman et al., 2004; Villani, Riva & Riva, 2007; Villani Attention to Response Task (SART). We also predicted
& Riva, 2008; Villani et al., 2009; Valtchanov et al., 2010). that the urban virtual environment would be more stressful
than the geometric environment since the urban virtual en-
Natural and urban environment types were used in order vironment contained features that may cause stress (such
to mirror previous literature on restorative effects in real as advertisements, various colored lights, and high levels
world studies (Ulrich, 1981; Ulrich et al., 1991; Sheets & of various types of noise), potentially elevating physio-
Manzer, 1991; Berman et al., 2008) which have compared logical stress and having a negative impact on affect and
exposure to natural environments to urban ones. The geo- sustained attention. Lastly, we predicted that immersion
metric environment was developed as a third comparison in the geometric environment would have no effect on all
in order to examine the question of whether nature envi- measures (relative to pre-immersion), since both major
ronments are neutral and urban environments stressful, or theories on restoration predicted results going in opposite
if nature environments are restorative and urban environ- direction (cancelling each other out) and the environment
ments neutral (Valtchanov et al, 2010). The current study contained neither threatening stimuli, nor anything resem-
aimed to address this question because much of the liter- bling nature (aside from colors). Thus it was used as the
ature on restorative effects of nature has only compared neutral control group.
natural versus urban environments, creating an ambiguity
as to whether nature is restorative or neutral since urban meThod
settings could be either neutral or stressful. parTiCipanTs
Prior to recruitment, participants were prescreened using
Furthermore, the geometric environment was developed a mass testing questionnaire. Participants were required to
to help test both the Attention Restoration Theory (ART) speak and read English fluently (in order to understand in-

JCR
362 Physiological and Affective Responses to VR

structions), to not have experienced seizures, vertigo, or ZIPERS (Zuckerman, 1977) was adapted from previous
motion sickness prior to the study (to reduce risk of sim- experiments exploring restorative effects of nature both in
ulator sickness during the study), and to have reported that the real world and within VR which have found it to be a
they have corrected-to-normal or normal vision. Partici- reliable measure of restoration (Ulrich, 1981; Ulrich et al.,
pants were also prescreened using the Witmer and Singer 1991; Hartig et al., 1991; Valtchanov et al., 2010).
(1998) Immersive Tendencies Questionnaire (ITQ). A ran- ZIPERS includes 12 items that measure positive affect
dom sample of 69 undergraduate students (32 male and (happiness, friendliness, playfulness, and affection), neg-
37 female), ages 18 to 26, who scored within one standard ative affect (anger, sadness, and avoidance) and attentive-
deviation of the mean on the ITQ were recruited to par- ness on a 5-point Likert scale. Two more items were
ticipate in the study in exchange for course credit. Partic- incorporated at the end of the questionnaire to measure
ipants were randomly assigned to one of the three self-reported stress. The two items were added at the end
conditions (nature, geometric or urban). There were 11 of the scale to prevent responses on them from influencing
males and 13 females (24 total) in the nature condition, responses on any of the original scale items that preceded
10 males and 12 females (22 total) in the geometric con- them.
dition, and 11 males and 12 females (23 total) in the urban
condition. Both skin conductance level (SCL) and heart rate (HR)
were continuously recorded by a computer from the be-
design ginning to the end of the experiment and were used as a
Measures of physiological, affective and cognitive states measure of physiological stress. Measurements were
were taken at the start of the experiment (baseline), post recorded using the ADInstruments PowerLab Data Acqui-
stress induction, and post immersion in VR. Each partici- sition System and accompanying LabChart software. Both
pant served as his or her own control on the repeated SCL and HR were recorded using a sampling rate of 100
measures. The VR experience required participants to ex- Hz. SCL was recorded using two fingertip electrodes at-
plore one of the three virtual environments, either the na- tached to the index and middle fingers of the participant’s
ture setting, the environment full of geometric shapes, or non-dominant hand. SCL is viewed as a measure of sym-
the urban setting. Procedures and measures were identical pathetic nervous system activity (Dawson, Schell & Fil-
between all three conditions. The only difference between ion, 2007; Valtchanov et al., 2010). Increases in
conditions was the virtual environment participants ex- sympathetic nervous system activity have been associated
plored. with increases in activation of epidermal tissue and sweat
glands, which results in secretion of sweat and an increase
measures in skin conductivity (Dawson, Schell & Filion, 2007;
Restorative effects were measured in three different ways Valtchanov et al., 2010). The tonic skin conductance com-
(self-report questionnaire, physiological recordings, and ponent (SCL) was used instead of the phasic component
behavioural task) in similar fashion to our previous work (skin conductance response: SCR) since SCL has been
(Valtchanov et al., 2010). ZIPERS (Zuckerman, 1977) was previously used in research on restorative effects of nature
used to measure positive and negative affect in accordance (de Kort et al., 2006; Valtchanov et al., 2010) and is asso-
with previous research by Ulrich et al. (1981;1991) and ciated with tonic states of sympathetic nervous system
Valtchanov et al. (2010). Measures of physiological stress arousal (e.g., stress) Dawson et al., 2007; Valtchanov et
consisting of skin conductance level and heart rate were al., 2010). HR was recorded using an infrared fingertip
used as they have been in previous research (de Kort et sensor placed on the ring finger of the participants’ non-
al., 2006; Villani et al., 2009; Valtchanov et al., 2010). To dominant hand, and served as a secondary measure of
measure “cognitive fatigue” the SART was employed, physiological stress. Both SCL and HR sensors and wires
which has been shown to be a sensitive behavioural meas- were secured to the participants’ hand using Velcro straps
ure of sustained attention ( Cheyne, Solman, Carrier, & to prevent sensors from moving during the experiment.
Smilek, 2009) and has previously been used to test indi-
viduals suffering from stress (Van Der Linden, Keijers, The third measure of restoration used was the SART. The
Eling, & Van Schaijk, 2005). Individuals suffering from SART is a powerful behavioral measure of sustained at-
(chronic) stress have been shown to have trouble using ex- tention that gives insight into some of the states of sus-
ecutive attention to inhibit the automatized response in the tained attention (Robertson et al., 1997; Manly et al., 1999,
SART (Van Der Linden, et al., (2005). Cheyne et al., 2009). The SART features a rapid display

JCR
Valtchanov and Ellard 363

of randomized numbers from 1 to 9. During the SART, sounds (such as the ocean, rivers flowing, etc.) were also
participants are instructed to press the “space” key on a present in the environment. The environment also featured
keyboard every time they see a number that is not the tar- simulated wind and realistic physics, creating a gentle
get, and are told to withhold their response (i.e., refrain sway of the tree branches and ripples/waves in the water.
from pressing the “space” key) when the target (#3) ap- It should be noted that participants did not physically feel
pears onscreen. This produces an automatized behavior of any wind or breeze inside the climate controlled lab room.
pressing the “space” key since eight out of nine responses The environment itself featured photorealistic graphics
require the key to be pressed. Attention must be sustained rendered in real-time using the CryEngine 2. This included
in order to prevent pressing the “space” key when the tar- (but was not limited to) things such as high dynamic range
get randomly appears. The main two measurements that (HDR) lighting, high resolution textures, and highly de-
emerge from the task are the number of inhibition errors, tailed foliage (that included the rendering of individual
where participants have failures of sustained attention and leaves, blades of grass, and flowers).
fail to use executive control to inhibit the automatized be-
haviour of pressing the “space key,” and response time. The geometric setting was created using an assortment of
Both inhibition errors and response time have been asso- 3-D geometric shapes including spheres, cylinders, cones,
ciated with sustained attention (Robertson et al., 1997; and rectangular and square boxes of various sizes. Shapes
Manly et al., 1999, Cheyne et al., 2009). were colored using the color palette found within the na-
ture environment (i.e., with a heavy focus on greens, blues,
Lastly, simulator sickness was measured using the Simu- and browns) in order to help control the potential effects
lator Sickness Questionnaire developed by Kennedy et al. of color on affect. The geometric environment did not fea-
(1993) which measures sickness induced by simulators ture any sounds apart from the individuals’ virtual foot-
(such as VR) using three subscales (nausea, oculomotor, steps since it was an empty environment (apart from the
disorientation). The questionnaire asks participants to rate shapes themselves).
the levels of various symptoms (nausea, blurred vision,
eyestrain, etc) that they may be experiencing on a scale The urban environment was a model (to scale) of the area
from “none” to “severe.” Simulator sickness was meas- surrounding Shibuya station in Tokyo, Japan. A hypothet-
ured in order to determine if measures of affect, sustained ical urban environment was not created since the environ-
attention, and stress were being confounded by the simu- ment was intended to capture modern design and
lator sickness that sometimes results from immersion in architecture accurately. Shibuya was chosen as the urban
VR (Kennedy et al., 1993; Valtchanov et al., 2010). environment because it is a dense urban center that par-
ticipants were unfamiliar with. It featured full-scale build-
TeChniCal informaTion abouT VirTual enViron- ings, streets, sidewalks with photorealistic HDR ambient
menTs and VirTual realiTy seTup lighting, and advertisements. All buildings, signs and ob-
Both the virtual nature environment and virtual urban en- jects were made using realistic textures taken from pho-
vironment were constructed using a combination of the tographs of the actual location in Japan. Ambient city
CryEngine 2 software developer’s kit, the Crysis level cre- sound was also present, giving the impression that people
ator, and Google Sketchup modelling software. The geo- were present but very far away. However, no actual 3-D
metric virtual environment was constructed using a models of people were present and there were no moving
combination of Google Sketchup and the Worldviz VR vehicles on the streets. The three environments (nature,
software, Vizard. All three environments were built to be geometric, and urban) can be seen in Figure 1.
of similar size and to allow a similar amount of explo-
ration on a grid of 3 km2. The overall layout of all three The three virtual environments were rendered using
environments was kept as consistent as possible while re- CryEngine 2, a platform available to individuals who have
taining the uniqueness of each type of virtual environment. purchased the PC game Crysis. Environments were ren-
The nature island contained two waterfalls (with accom- dered using a consumer-grade gaming PC with a 2.4 GHz
panying rivers flowing toward the ocean), various kinds quad-core processor, 4 GB of ram and two AMD ATI
of palm trees and types of broad-leaved trees, grass, rocks, 3870x2 video cards at 1280x1024 pixels per eye with pho-
varied flower bushes and plants, and a long stretch of torealistic shadows and lighting. The rendered scene was
beach by the ocean. Dirt path networks were clear and then piped in stereo to an nVIS head-mounted display
easy to follow throughout the island. Ambient nature (HMD) that featured a 65-degree field of view. The HMD

JCR
364 Physiological and Affective Responses to VR

also featured a thick light blocking cover which prevented which a SCL and HR baseline was established for 2 min.
participants from seeing the real world environment the SART was then administered on the computer for 5
around them, allowing them to better focus on the virtual min in order to establish baseline performance and ability
environment. to sustain attention.

The viewpoint was controlled by an InterSense Inerti- Participants were then given a 10 min stress induction task
aCube2 head-tracking device which was attached to the on the computer consisting of a modified version of the
HMD, allowing the viewpoint to update in real-time with Markus & Peters Arithmetic Test (previously found to be
physical head movements (i.e., if a participant turned their effective at stress induction by de Kort et al. (2006)). Dur-
head toward the right, the viewpoint in VR would update ing the task, participants were asked to use mental arith-
and look toward the right.) Self-locomotion through the metic to solve difficult multi-step multiplication questions
virtual environments was achieved through the use of a (e.g., 56 x 37 + 17) within a time limit (a 60 sec timer was
wireless mouse calibrated to move participants’ “virtual present for each question). While solving the questions,
bodies” forward and backward with the left and right click participants wore stereo headphones that played loud
respectively. The direction of movement corresponded to street traffic noises meant to distract them and make the
the orientation of their head in the physical world. Partic- task more difficult and frustrating. Participants were also
ipants were instructed to orient their body to the direction given feedback after every incorrect response to induce
of movement in order to reduce the mismatch between further stress. A homogenized stress induction was used
their physical and virtual bodies. to induce stress in participants within the controlled lab
setting in order bring participants to a similar level of
proCedure stress and negative affect prior to immersion in VR. This
Participants were randomly assigned to one of the three was done so that the effectiveness of each virtual environ-
conditions (nature, geometric, or urban). An identical pro- ment, in helping participants to recover from a similar
cedure was followed for all three conditions with the ex- level of stress, could be measured, and between-group
ception of the type of virtual environment (nature, comparisons could be made.
geometric, or urban) participants were exposed to. Each
participant was met in a central meeting area by appoint- After the stress induction, the SART was administered
ment and was then escorted to a lab setting where he or again for 5 min to measure if participants’ ability to sustain
she participated in the experiment. Participants were attention had changed from baseline. After administration
given a cover story stating that the purpose of the experi- of the SART, participants filled out ZIPERS for 2 min once
ment was to examine the effect of stress and VR on their again while SCL and HR data was collected. The SCL and
performance on the SART. HR data file was marked with the start and finish of the 2
min interval. ZIPERS scores and SCL and HR data was
At the start of the experiment, participants were seated at collected post stress and at the administration of the SART
a desk in front of a computer monitor with a keyboard and in order to have an accurate measure of participants’ phys-
mouse, and were hooked up to SCL and HR sensors on iological and affective states just prior to exploring VR.
their non-dominant hand in order to measure physiological
stress levels. Participants wore the SCL and HR sensors After participants finished filling out ZIPERS, they were
on their non-dominant hand throughout the entire experi- fitted with a HMD (with motion tracking) immersing them
ment and were asked to keep the hand stationary on the in one of the three virtual environments (nature, geometric,
desk while they filled out questionnaires (in order to pre- or urban) seen in Figure 1, based on random assignment.
vent measurement noise and movement artifacts). In order Participants were given 10 min to freely explore the virtual
to acclimatize participants to the equipment prior to ob- environment, with their only goal being to find objects or
taining measurements, participants wore the SCL and HR locations which they found pleasant or interesting.
sensors for 5 min while sitting at the desk during which
the experiment protocol was explained. They were then After the 10 min VR session was finished, participants
asked to fill out ZIPERS on the computer using the key- were once again given ZIPERS for 2 min to measure the
board and their dominant hand in order to establish a base- effects of the virtual environments on their emotional state
line for self-reported affective state (on levels of positive while SCL and HR were once again collected and
affect, negative affect, ability to focus and stress) during recorded. Finally, the SART was administered for 5 min

JCR
Valtchanov and Ellard 365

with the self-report of feeling stressed. As seen in Table

1, SCL increased significantly F(1,68) > 100, p < 0.001,
MSE = 0.621, ηp2 = 0.796, and HR trended increase,
F(1,68) = 2.771, p = 0.10, MSE = 39.577, ηp2 = 0.039.
Figure 1. Screen captures of the three virtual environ- These results suggest that the stress induction was suc-
ments with the nature environment on the left, the geo- cessful at inducing negative mood, and both self-reported
metric environment in the middle and the urban and physiological stress.
environment on the right. The nature environment fea-
tured a photorealistic nature island that was fully ex- A one-way ANOVA was used to compare group mean
plorable and was surrounded by an ocean. The scores post stress induction to test if the stress induction
geometric setting featured solid colored geometric prompted all three groups to experience a similar level of
shapes comprised of boxes, cones and cylinders. stress and affect. As expected, no significant differences
Lastly, the urban environment featured a photorealistic in positive affect scores were found between the nature
recreation of the area around Shibuya Station in Japan. (M = 2.208, SD = 0.714), urban (M = 2.400, SD = 0.826)
and geometric (M = 2.391, SD = 0.670) groups, F(2,66) =
for the final time to measure any changes in ability to sus- 0.505, p = 0.61, n.s., MSE = 0.544, η2 = 0.015. No sig-
tain attention resulting from immersion in VR. nificant differences in negative affect scores were found
between the nature (M = 1.883, SD = 0.854), urban (M =
resulTs 1.555, SD = 0.611) and geometric (M =1.504, SD = 0.655)
reliabiliTy analysis of Zipers groups, F(2,66) = 1.932, p = 0.15, n.s., MSE = 0.516, η2
To simplify data analysis, the questions measuring happi- = 0.055. No significant differences in self-reported atten-
ness, friendliness, affection, and playfulness were grouped tiveness scores were found between the nature (M = 2.625,
under the broad category of “positive affect.” These items SD = 1.173), urban (M = 2.682, SD = 1.086) and geomet-
were found to have an acceptable Cronbach’s alpha of ric (M = 2.826, SD = 1.114) groups, F(2,66) = 0.197. p =
0.88. Similarly, the questions measuring fear, anger, and 0.821, n.s., MSE = 1.268, η2 = 0.006. Also, there were no
sadness were grouped into the broad category of “negative significant differences in self-reported stress between the
affect.” These items were also found to have an acceptable nature (M = 3.833, SD = 1.167), urban (M = 3.955, SD =
Cronbach’s alpha of 0.81. 0.950) and geometric (M = 3.783, SD = 0.736) groups,
F(2,66) = 0.186, p = 0.83, n.s., MSE = 0.942, η2 = 0.006.
sTress manipulaTion CheCk As seen in Table 1, there were also no significant differ-
To confirm that the stress induction was successful and ences between the nature, urban, and geometric groups
that participants were stressed and experiencing a nega- post stress induction (prior to VR immersion) in SCL,
tive emotional state prior to their VR experience such that F(2,66) = 2.034, p = 0.14, n.s., MSE = 1.205, η2 = 0.058,
restoration could take place, baseline scores on ZIPERS, or in HR, F(2,66) = 1.552, p = 0.22, n.s., MSE = 139.859,
SCL, and HR levels obtained prior to stress induction and η2 = 0.045. These results confirm that after the stress in-
post stress induction were compared using a set of mixed- duction, all three groups were experiencing a similar level
model repeated measures ANOVAs. As a result of the of stress and affect and that there were no significant dif-
stress induction, participants reported feeling significantly ferences in affect or stress between groups.
lower positive affect (M = 2.3, SD = 0.73) than at baseline
(M = 2.8, SD = 0.70), F(1,68) = 32.90, p < 0.001, MSE = However, the stress induction failed to induce cognitive
0.234, ηp2 =0.326. Participants also reported a significant fatigue and lapses in sustained attention, since there was
increase in negative affect (M = 1.65, SD = 0.73) com- no significant change in the SART inhibitory errors,
pared to baseline (M =1.29, SD = 0.42), F(1,68) = 18.90, F(1,68) = 0.072, p = 0.789, n.s., MSE = 3.599, ηp2 =
p < 0.001, MSE = 0.228, ηp2 = 0.217, significantly higher 0.001, or SART response time, F(1,68) = 1.085, p = 0.301,
feelings of stress (M = 3.85, SD = 0.96) compared to n.s., MSE = 1281.444, ηp2 = 0.016. This suggests that par-
baseline (M = 2.97, SD = 1.2), F(1,68) = 37.78, p < 0.001, ticipants’ ability to sustain attention remained at baseline
MSE = 0.714, ηp2 = 0.357, and significantly lower atten- despite elevated levels of stress and negative affect.
tiveness (M = 2.7, SD = 1.11) compared to baseline (M =
3.28, SD = 0.89), F(1,68) = 12.39, p < 0.001, MSE = effeCTs of Vr on Zipers sCores
0.889, ηp2 = 0.154. Physiological measures converged Scores on the Simulator Sickness Questionnaire (SSQ)

JCR
366 Physiological and Affective Responses to VR

were used to check if participants were ex-

periencing any simulator sickness which
may have influenced scores. Participants
reported feeling “none” (1) to “slight” (2)
simulator sickness on a scale from 1 to 4
(M=1.645, SD = 0.456), thus simulator
sickness was not believed to be a confound.
To examine the effects of immersion into
VR on participants’ self-reported affect,
ZIPERS scores obtained just prior to im-
mersion into VR (but post stress induction)
were compared to scores obtained imme-
diately after immersion into VR, using a set
of mixed-model repeated measures
ANOVAs with time (prior to VR, post VR)
as the repeated measure and condition (na-
ture, urban, geometric environments) as the
between variable. Figure 2. Mean positive affect scores (with standard error bars) on the
Zuckerman Inventory of Personal Reactions Scale (ZIPERS) per con-
A significant time by condition interaction dition at baseline, post stress induction, and then post immersion in
on the positive affect dependent variable VR. Here it can be seen that positive affect decreased in all three
was found, F(2,66) = 9.676, p < 0.001, groups as a result of the stress induction (as seen by comparing the
MSE = 0.328, ηp2 = 0.227, suggesting that first and second bars in each set), and only improved when partici-
the three virtual environments had differ- pants explored a virtual nature environment (seen by comparing the
ent effects on positive affect. To test for second and third bars in each set).
simple effects, the data was split by condi-
tion, and repeated measures ANOVAs
were done. The analysis revealed that the
nature virtual environment caused positive
affect to increase significantly from a
mean of 2.21 (SD = 0.71) prior to VR to a
mean of 3.03 (SD = 0.98) post VR, F(1,23)
= 14.304, p < 0.001, MSE = 0.571, ηp2 =
0.383. Meanwhile the geometric virtual
environment had no effect (M = 2.39, SD
= 0.67 prior to VR to M = 2.27, SD = 0.94
post VR), F(1,22) = 0.776, p = 0.388, n.s.,
MSE = 0.220 ηp2 = 0.034, and the urban
environment had no effect (M = 2.40, SD
= 0.83 prior to VR to M = 2.38, SD = 0.74
post VR), F(1,21) = 0.021, p = 0.887, n.s.,
MSE = 0.175, ηp2 = 0.001. The means and
standard error bars for each of the condi- Figure 3. Mean negative affect scores (with standard error bars) on
tions can be seen in Figure 2. ZIPERS per condition at baseline, post stress induction, and then post
immersion in VR. Here it can be seen that there was a potential floor
A significant time by condition interaction effect since all scores were close to the bottom of the scale throughout
on the negative affect dependent variable the experiment.
was also found, F(2,66) = 6.140, p < 0.005,
MSE = 0.180, ηp2 = 0.157, suggesting that
the virtual environments had different ef-

JCR
Valtchanov and Ellard 367

fects on participants’ level of negative af-

fect. To test for simple effects, the data was
split by condition and repeated measures
ANOVAs were used. The analysis re-
vealed that the nature virtual environment
caused negative affect to decrease signifi-
cantly from a mean of 1.88 (SD = 0.85)
prior to VR, to a mean of 1.22 (SD = 0.35)
post VR, F(1,23) = 17.750, p < 0.001,
MSE = 0.293, ηp2 = 0.436. The geometric
environment had no effect (M = 1.50, SD
= 0.65 prior to VR to M = 1.45, SD = 0.62
post VR), F(1,22) = 0.268, p = 0.610, n.s.,
MSE = 0.117, ηp2 = 0.012, while the
urban environment caused a significant de-
crease in negative affect from a mean of
1.55 (SD = 0.61) prior to VR, to a mean of
1.27 (SD = 0.41) post VR, F(1,21) = 7.205, Figure 4. Mean stress self-report scores on a 1 to 5 scale with standard
p = 0.014, MSE = 0.121, ηp2 = 0.255. The error bars. Here it can be seen that the stress induction was successful
means and standard error bars for each of in bringing participants in all three conditions to the same level of
the conditions can be seen in Figure 3, stress prior to immersion in VR (middle bar), confirming that partici-
where it also appears that there might be a pants were similarly stressed between conditions before exploring VR.
floor effect since scores are near the lower Here it can also be seen that participants who explored the nature vir-
end of the scale. tual environment had the greatest decrease in stress (seen by compar-
ing the second and third bars in each set).
A significant time by condition interaction
on self-reported stress was found, F(2,66)
= 12.763, p < 0.001, MSE = 0.918, ηp2 =
0.113, converging with previous measures
and suggesting that the three virtual envi-
ronments had different effects on per-
ceived stress levels. Simple effects
analysis using repeated measures
ANOVAs revealed that self-reported stress
decreased significantly from a mean of
3.83 (SD = 1.17) to a mean of 2.63 (SD =
1.35) as a result of being immersed in the
nature environment, F(1,23) = 12.219, p <
0.005, MSE = 1.434, ηp2 = 0.347, but did
not change significantly as a result of
being immersed in the geometric environ-
ment (M = 3.78, SD = 0.74 prior to VR to
M = 3.70, SD = 1.11 post VR), F(1,22) =
Figure 5. Mean attentiveness score (with error bars) on ZIPERS per
0.193, p = 0.665, n.s., MSE = 0.451, ηp2=
condition at baseline, post stress induction, and then post immersion in
0.009, or the urban environment (M =
VR. Here it can be seen (by comparing the first and second bars in
3.95, SD = 0.95 prior to VR to 3.50, SD =
each set) that self-reported attentiveness decreased in all three condi-
0.96 post VR), F(1,21) = 2.692, p = 0.116,
tions as a result of the stress induction. It can also be seen (by compar-
n.s., MSE = 0.844, ηp2 = 0.114. The
ing the second and third bars in each set) that attentiveness stopped
means and standard error bars for each of
declining after participants explored the nature environment, but not
the conditions can be seen in Figure 4.
when they explored either the geometric or urban environments.

JCR
368 Physiological and Affective Responses to VR

Table 1
Means and Standard Deviations of Physiological Measures

Physiological
Condition Baseline Post-stress induction Post VR immersion
measure

Nature M=82.91 M=86.16 M=78.22

SD=12.92 SD=11.46 SD=13.42
Geometric M=81.77 M=82.40 M=76.97
Heart rate
SD=14.57 SD=11.33 SD=10.12
Urban M=78.67 M=80.08 M=76.78
SD=12.66 SD=12.70 SD=12.39

Nature M=2.50 M=1.76

-
SD=0.98 SD=1.14
Skin conductance
Geometric M=1.86 M=2.02
level change from -
SD=0.95 SD=1.40
baseline (z-score)
Urban M=2.15 M=2.26
-
SD=1.33 SD=1.37

Lastly, a significant time by condition interaction on ures, measuring the physiological state of participants
the self-reported attentiveness dependent variable was as they were reflecting and filling out the self-report.
also found, F(2,66) = 3.436, p < 0.05, MSE = 0.638,
η p2 = 0.094. Simple effects were once again tested For the purpose of analysis, skin conductance measure-
using repeated measures ANOVAs. The analysis re- ments were converted into z-scores to standardize SCL re-
vealed that self-reported attentiveness (i.e., ability to sponses between participants, so that between-subject
focus) did not change as a result of immersion in the comparisons could be made. This was required in order to
virtual nature environment (M = 2.63, SD = 1.17 prior prevent bias in the dataset due to “high-responding” indi-
to VR to M = 2.67, SD = 0.87 post VR), F(1,23) = viduals whose changes in SCL magnitudes were up to 10
0.038, p = 0.846, MSE = 0.543, ηp2 = 0.002, but de- times the average. Skin conductance was measured rela-
creased significantly from a mean of 2.83 (SD = 1.11) tive to a baseline established at the start of the experiment.
to a mean of 2.04 (SD = 0.93) as a result of being in the The equipment was zeroed at this baseline. Similarly to
geometric virtual environment, F(1,22) = 11.960, p < the analysis of the self-report data above, measurements
0.005, MSE = 0.589, ηp2 = 0.352. Self-reported atten- prior to immersion in VR were compared to measurements
tiveness also decreases significantly from a mean of post VR immersion using a mixed-model repeated meas-
2.68 (SD = 1.09) to a mean of (2.09, SD = 0.75) as a ures ANOVA. A significant time by condition interaction
result of being in the urban virtual environment, F(1,21) was found on the dependent variable of SCL, F(2,66) =
= 4.842, MSE = 0.793, ηp2 = 0.187. This data can be 7.166, p < 0.005, MSE = 0.424, ηp2 = 0.178, suggesting
seen in Figure 5. that the three virtual environments had different effects on
participants’ stress levels. Simple effects analyses were
sTaTisTiCal analysis of The sCl and hr measures done by splitting the data by condition and performing re-
Both SCL and HR were recorded continuously through- peated measures ANOVAs. The analyses revealed that (as
out the experiment. The data file was marked with the seen in Table 1) SCL decreased significantly as a result of
start and finish of each ZIPERS questionnaire that was immersion in the nature environment, F(1,23) = 11.625, p
administered (i.e., at baseline, post stress induction, and < 0.005, MSE = 0.577, ηp2 = 0.336, but did not change
post VR immersion). SCL and HR were averaged over as a result of immersion in the geometric environment,
the 2 min intervals during which ZIPERS was admin- F(1,22) = 0.617, p = 0.441, n.s., MSE = 0.426, ηp2 =
istered so that the congruent physiological (stress) state 0.027, or in the urban environment, F(1,21) = 0.557, p =
could be captured and potential movement artifacts 0.464, n.s., MSE = 0.254, ηp2 = 0.026, converging with
could be diminished. This ensured that physiological the self-report stress measure. Means and standard devia-
measures were temporally linked with self-report meas- tions can be seen in Table 1.

JCR
Valtchanov and Ellard 369

HR was also analyzed using a mixed-model repeated puter-generated nature stimuli produce restoration when
measures ANOVA. The analysis revealed a trending time participants are immersed in VR. Immersion in the virtual
by condition interaction on the dependent variable of HR, nature setting prompted an increase in positive affect (hap-
F(2,66) = 2.997, p = 0.057, MSE = 20.704, ηp2 = 0.083. piness, friendliness, affection, and playfulness), and a de-
Simple effects analyses revealed that HR decreased sig- crease in negative affect (fear, anger, sadness) as seen in
nificantly as a result of immersion in the nature environ- Figures 2 and 3. Furthermore, a significant self-reported
ment, F(1,23) = 25.842, p < 0.001, MSE = 29.242, ηp2 = decrease in levels of perceived stress (seen in Figure 5)
0.529, the geometric environment, F(1,22) = 34.714, MSE converged with the significant decrease in levels of phys-
= 9.754, p < 0.001, ηp2 = 0.612, and the urban environ- iological stress (as measured by SCL). These results are
ment, F(1,21) = 5.242, p < 0.05, MSE = 22.824, ηp2 = consistent with findings by Valtchanov et al. (2010) who
0.200, suggesting that HR decreased significantly regard- have previously reported that virtual nature can have
less of the virtual environment as seen in Table 1. restorative properties on affect and physiological stress,
Statistical analysis of the SART measure as well as research by de Kort et al. (2006) and van den
Berg et al. (2003) which suggests that surrogate nature
A set of mixed-model repeated measures ANOVAs were (such as photographs and videos) can have similar restora-
used to analyze the rate of inhibition errors and response tive effects.
time on the SART prior to immersion in VR (post stress
induction) and post immersion in VR. The analyses re- Secondly, it becomes much more apparent that it is virtual
vealed no significant or trending interactions of time by nature that is responsible for the observed restoration and
condition on the dependent variable of inhibitory error not VR itself when the group that was immersed in the vir-
rate, F(2,66) = 1.607, p = 0.176, n.s., MSE = 3.470, ηp2 = tual nature setting is compared to the group immersed in
0.024, or the dependent variable of reaction time, F(2,66) the geometric setting, and the group immersed in the urban
= 0.203, p = 0.817, n.s., MSE = 1182.581, ηp2 = 0.006. setting. All participants underwent an identical protocol
This supports the notion that participants were performing with the exception of the virtual environment they ex-
at baseline since the stress induction failed to induce cog- plored. All three virtual environments were equally large
nitive fatigue, thus no restoration could occur. and afforded similar exploration paths, while providing ap-
propriate cues (e.g. being able to hear the water near the
disCussion ocean, or being able to hear the train near the train station).
In the current study, a variety of measures of restorative All three environments also had a similar level of interac-
effects were used, including a standardized self-report tivity (i.e., participants could explore freely and look at
questionnaire measuring affect, perceived attentiveness whatever they wanted). Despite the similarity of the VR
and stress, physiological measures of SCL (stress) and experiences and identical protocol, only those who were
HR, and a behavioral measure of sustained attention immersed in virtual nature experienced a significant and
(SART) in order to thoroughly document the effects of consistent improvement of affect and reductions in stress
three unique VR environments on affect, stress, and at- across all measures, while those immersed in the geometric
tention. In doing so, the current study was able to pro- and urban environments displayed either no improvement
duce similar restorative effects to those reported by or ambiguous improvement (as shown by reductions in HR
Valtchanov et al. (2010) when immersing participants in in the absence of decreases in SCL and self-reported
virtual nature, while also addressing several other impor- stress). These findings support our predictions and suggest
tant questions: Is it possible to determine whether it was that the content of the VR experience, specifically the pres-
the nature within VR, or VR itself that was responsible ence of nature, and not the VR experience itself is likely
for reductions in stress and improvement in affect? The responsible for the observed restoration. This notion is
study also allowed determining whether the commonly supported by the previously documented restorative prop-
used urban comparison group (Ulrich, 1981; Ulrich et al., erties of exposure to nature (Berman et al., 2008; Kweon
1991; Sheets & Manzer, 1991; Berman et al., 2008) was et al., 2008, Valtchanov et al., 2010).
neutral or stressful in terms of eliciting an affective and
physiological response. When exploring whether urban settings were stressful (as
predicted) or neutral, we were able to come to a context-
Firstly, the current study found converging evidence from specific conclusion. The results of the current study sug-
several measures that supported our hypothesis that com- gest that in the absence of threatening stimuli (such as

JCR
370 Physiological and Affective Responses to VR

moving cars, cyclists, and other pedestrians) that are nor- 2001), nor Ulrich et al.’s (1981;1991) theories can fully
mally present in an urban setting, (requiring individuals explain. In the case of the decrease of negative affect
to be vigilant to avoid collisions or other threats,) virtual after exposure to the urban virtual setting, ART cannot
urban settings may reduce negative affect (anger, sadness, provide consistent explanation – since the environment
and fear), but do not improve positive affect or reduce did not contain threatening stimuli that demanded atten-
stress. We believe this to be a characteristic of benign tion (such as moving vehicles), it could be argued that
urban virtual settings since a similar effect was not present directed attention mechanisms were situated in a setting
in the group that explored the geometric virtual environ- where disengagement was relatively easy. Once partic-
ment. The geometric virtual environment proved consis- ipants learned that there were no threats within the envi-
tently to have no effect on stress and affect (as predicted) ronment, recovery from a negative emotional state
and was thus used as the neutral comparison. The promo- ensued. However, all participants were explicitly told
tion of decreases in negative affect by urban settings in the that there were no threats within the virtual environment
absence of other changes may have been the result of the and that harm within the virtual world was impossible,
environment distracting individuals from their awareness so such an explanation seems inaccurate. Similarly, it
of their negative emotions. In doing so, it could have in- could also be argued that participants habituated to the
fluenced perceived affect (measured by the self-report) (potentially) “attention demanding” advertisements part-
without influencing physiological state. This was not seen way through their exploration, which then allowed di-
in the geometric environment since the urban environment rected attention mechanisms to disengage and recover.
may have been more “attention grabbing” (given that it This explanation also seems inaccurate since the geomet-
contained many different advertisements while the geo- ric environment, which was similar in structural features
metric environment only contained solid colored shapes). to the urban environment but did not contain any of the
However, this cannot be said for certain as it was not di- advertisements, did not show the effect. Ulrich et al.’s
rectly measured. It should be noted that the effects ob- theory also fails to explain the observed pattern of results
served in this study were all in the context of virtual for the urban and geometric environments. In the current
environments that lacked threatening stimuli. It is likely study, participants had an affective response to the stim-
that the restorative effects of nature would disappear if the ulus (the urban environment) but did not exhibit a con-
nature environment was filled with threatening stimuli gruent physiological response as Ulrich et al.’s theory
(such as dangerous animals). Another important note is predicts. Furthermore, the observed affective and phys-
that the previous study by Valtchanov et al. (2010) and the iological responses are the reverse of the theoretical pro-
current study both used nature settings with lush vegetation jections. Since both environments did not contain any
(trees, flowers, bushes). It is possible that the restorative nature stimuli, Ulrich et al.’s theory predicts that they
effects of virtual nature may be restricted to natural settings would be perceived as threatening and would increase
containing dense vegetation. Further research is needed to stress and cause a deterioration of affect. However, nei-
clarify if other types of nature environments (e.g., rocky ther of the environments elicited a change in self-re-
terrain, mountains, snow-covered landscapes, grassy fields) ported or measured physiological stress, and negative
can also promote restoration. For example, Heerwagen affect decreased due to exposure to the urban environ-
and Orians (1992) suggest that our ancestors evolved to ment.
prefer habitats that offered access to food, water, and shel-
ter from predators and the elements, which increased their Given the presented data, several criticisms can be made
chances of survival. Heerwagen and Orians further suggest about both ART as proposed by Kaplan (1995;2001) and
that properties of good habitats became associated with Ulrich et al.’s (1981;1991) “affective response” theory.
positive emotional states during the evolutionary process. First and foremost, ART is ambiguous about what it means
Thus, it is possible that various types of natural environ- for a stimulus to modestly capture or demand attention,
ments that have the characteristics of a “good habitat” may which makes it difficult to test empirically. However, de-
promote positive emotional states, and even restoration. spite this ambiguity, the theory still fails to encompass the
results of the current study, predicting contradictory results
The unpredicted decrease in negative affect due to expo- to those observed. Similarly, Ulrich et al.’s (1981;1991)
sure to the urban setting, and the predicted and observed evolutionarily based “affective response” theory also fails
neutrality of geometric environment, present a conun- to predict the results observed in the current study. The
drum that neither Kaplan’s ART (Kaplan, 1995; Kaplan, presented data not only (partially) discredits both theories,

JCR
Valtchanov and Ellard 371

but also demonstrates that a new theory is required to ex- ence in their structural and geometric properties. Urban
plain the effects on affect and stress that the three virtual settings can be consistently described in terms of regular
environments promoted. From a technical perspective, all geometric shapes, while nature scenes contain signifi-
three VR experiences were identical with the exception of cantly more complex structural properties that lack a dis-
the visual and auditory information presented, yet partic- crete geometric definition. Given this, it is certainly
ipants had significantly different responses to the three vir- feasible that on a lower perceptual level, different neural
tual environments. While the visual and aural information networks are activated by nature versus urban stimuli, re-
can be identified as the cause of the differences in affect sulting in an automatic psychophysiological response.
and stress elicited between the nature and urban environ-
ments, it cannot fully explain why there were only minor Unfortunately, the current study did not offer converging
or insignificant differences between participants exposed evidence for the effects of VR on the ability to sustain at-
to the urban and geometric environments. The urban en- tention, since the stress induction manipulation in the cur-
vironment contained visual and auditory stimuli (adver- rent study failed to exhaust the cognitive resources of
tisements, sounds of crowds and vehicles in the distance, participants such that performance on the SART would
etc.) which restoration theories predict to be stressful, yet deteriorate. Participants were performing at baseline lev-
this environment was no more stressful than the geometric els on the SART before, and after stress induction, and
environment. The similarity between the effects of the after immersion in VR. However, it is interesting to note
geometric and urban environments on stress and affect that while there were no changes in SART performance,
suggests that the stimuli responsible for nature’s restora- participants who explored the nature environment reported
tive properties may be nested in the structural properties no perceived deterioration of attentiveness, whereas those
of the visual as opposed to auditory information. In sup- who explored the geometric and urban environments sub-
port, it can be argued that the visual properties of the urban jectively reported a significant decrease in perceived at-
and geometric environments were similar, but quite dis- tentiveness. While interesting, this finding is not supported
tinct from the nature environment. For example, a by the behavioral data from the SART (i.e., if participants
cityscape is merely a composite of regular geometric were truly less attentive, they should have made more er-
shapes (cubes, spheres, cones, cylinders, etc.) once color rors on the SART). It is possible that the SART was not
and texture are removed from the visual information. Co- sensitive enough to measure the deterioration, but it is also
incidentally, the geometric environment used in this ex- possible that participants subjectively perceived deterio-
periment matches such a description (but includes colors ration of attention. A correction for further research would
seen within nature, which eliminates color as a potential be to extend the stress induction to 20 or 30 min to prop-
stimulus for restoration). Meanwhile, once texture and erly induce cognitive fatigue.
color are removed from the nature environment, discrete
geometric shapes are absent. impliCaTions for fuTure researCh and ConClusion
The current study provides evidence that computer-gen-
Thus, it is plausible that the structural information of the erated nature can be used to reduce both perceived and
visual stimuli (such as the pattern of shapes, which can be physiological levels of stress, as well as improving posi-
further decomposed into spatial frequencies) may hold the tive affect (happiness, friendliness, playfulness) and re-
key to discovering what makes natural patterns restorative. ducing negative affect (fear, anger, and sadness). In
Such a notion is supported by examining the early work demonstrating that virtual nature has the power to pro-
of Ulrich (1981), and later work by Kweon et al. (2008) mote restoration (compared to other virtual urban scenes)
which demonstrated that viewing photographs of nature we propose that some of the benefits of exposure to nature
promotes improved psychophysical states when compared (specifically reductions in stress and improvement of af-
to viewing photographs of urban settings or abstract fect), can be harnessed using technology and brought into
posters. In these scenarios, only the visual information the global urban culture that has pushed many individuals
differed, yet nature images consistently promoted more into densely populated urban centers, away from nature.
restoration despite being photographs of different nature Future research aimed at stress management and clinical
settings, such as water, grass, trees, and mountains (thus uses of VR to improve quality of life should focus on ex-
containing different visual information). Given this pat- ploring ways to use nature stimuli to augment treatments.
tern, or lack thereof, one can begin to see that the major Since the current study was able to generate an effective
difference between nature and urban stimuli is the differ- restorative environment using a consumer-grade com-

JCR
372 Physiological and Affective Responses to VR

puter and toolkits available to the general public, we be- into their living room. Thus, we encourage further re-
lieve that researchers should begin to directly target ther- search into how these technologies can be used to bring
apeutic solutions that can be brought to the general better mental health into the homes of the general public.
consumer market. Raw computing power and software Lastly, research aimed at exploring “why” nature stimuli
have become relatively inexpensive in the last decade, promote restorative effects in individuals should focus
and with the rapid advances in technology in the field of on examining the structural properties of nature versus
3-D televisions, computer monitors, projectors and urban and abstract stimuli, as these may hold the key to
screens, and consumer-grade head-mounted displays, the discerning the qualities of nature stimuli that make them
general consumer market is at the brink of bringing VR restorative.

referenCes Orians, G.H., & Heerwagen, J. (1992). Evolved responses

Cheyne, J., Solman, G. J. F., Carriere, J. S. A., & Smilek, to landscapes. In The adapted mind: Evolution-
D. (2009). Anatomy of an error: A bidirectional ary psychology and the generation of culture,
state model of task engagement/disengagement edited by Barkow, J., Cosmides, L., & Tooby,
and attention related errors. Cognition, 111(1), J. Oxford and New York: Oxford Univeristy
98-113. Press.

Berman, M. G., Jonides, J., & Kaplan, S. (2008). The cog- Kaplan, S. (1995). The restorative benefits of nature: To-
nitive benefits of interacting with nature. Psy- ward an integrative framework. Journal of En-
chological Science, 19(12), 1207-1212. vironmental Psychology, 15,169–182.

Dawson ME, Schell AM, Filion DL. (2007) The electro- Kaplan, S. (2001). Meditation, restoration, and the man-
dermal system. In: Cacioppo JT, Tassinary LG, agement of mental fatigue. Environment and
Bernston GG, eds. Handbook of psychophysi- Behavior, 33, 480–506.
ology. 3rd ed. New York: Cambridge Univer-
sity Press, pp. 159–81. Kennedy, R.S., Lane, N.E., Berbaum, K.S., & Lilienthal,
M.G. Simulator sickness questionnaire: an en-
de Kort, Y. A. W., Meijnders, A. L., Sponselee, A. A. hanced method for quantifying simulator sick-
G., & Ijsselsteijn, W. A. (2006). What's ness. International Journal of Aviation
wrong with virtual trees? Restoring from Psychology, 3(3):203-220, 1993.
stress in a mediated environment. Journal
of Environmental Psychology, 26(4), 309- Kweon, B., Ulrich, R. S., Walker, V. D., & Tassinary, L.
320. G. (2008). Anger and stress: The role of land-
scape posters in an office setting. Environment
Freeman, J., Lessiter, J., Keogh, E., Bond, F. W., & Chap- and Behavior, 40(3), 355-355.
man, K. (2004). Relaxation Island: Virtual, and
really relaxing. 7th International Workshop on Leather, P., Pyrgas, M., Beale, D., & Lawrence, C. (1998).
Presence/Presence 2004, 13–15 October, Uni- Windows in the workplace: Sunlight, view, and
versitat Politecnica de Valencia, Spain. occupational stress. Environment and Behavior,
30(6), 739-762.
Gullone, E. (2000). The Biophilia hypothesis in the 21st
century: Increasing mental health or increasing Manly, T., Robertson, I. H., Galloway, M., & Hawkins, K.
pathology? Journal of Happiness Studies, 1(3), (1999). The absent mind: Further investigations
293-321. of sustained attention to response. Neuropsy-
chologia, 37, 661–670.
Hartig, T., Mang, M., & Evans, G. W. (1991). Restorative
effects of natural environment experiences. En- Peters, M. L., Godaert, G. L., Ballieux, R. E., van Vliet,
vironment and Behavior, 23(1), 3-26. M., Willemsen, J.J., Sweep, F. C., et al. (1998).

JCR
 Valtchanov and Ellard 373

Cardiovascular and endocrine responses to ex- Van Den Berg A.E., Koole S.L., van der Wulp N.Y.
perimental stress: Effects of mental effort and (2003). Environment preference and restora-
controllability. Psychoneuro-endocrinology, tion: (How) are they related? Journal of Envi-
23(1), 1–17. ronmental Psychology. 23(2), 135-146.

Robertson, I. H., Manly, T., Andrade, J., Baddeley, B. T., Van Der Linden, D., Keijsers, G. P. J., Eling, P., & Van
& Yiend, J. (1997). ‘Oops!:’ Performance cor- Schaijk, R. (2005). Work stress and atten-
relates of everyday attentional failures in trau- tional difficulties: An initial study on burnout
matic brain injured and normal subjects. and cognitive failures. Work and Stress,
Neuropsychologia, 35(6), 747–758. 19(1), 23-36.

Sheets, V. L., & Manzer, C. D. (1991). Affect, cognition, Villani, D., Riva, F., & Riva, G. (2007). New technologies
and urban vegetation: Some effects of adding for relaxation: The role of presence. Interna-
trees along city streets. Environment and Be- tional Journal of Stress Managemen, 14(3),
havior, 23(3), 285-304. 260-274.

Ulrich, R. S. (1981). Natural versus urban scenes: Some Villani, D., & Riva, G. (2008). The role of media in sup-
psychophysiological effects. Environment and porting a stress management protocol: An ex-
Behavior, 13(5), 523-556. perimental study. Journal of Cybertherapy and
Rehabilitation. 1(2), 159-173.
Ulrich, R.S., Simons, R.F., Losito, B.D., Fiorito, E., Miles,
M.A., & Zelson, M. (1991). Stress recovery Villani, D., Luchetta, M., Preziosa A., & Riva, G. (2009).
during exposure to natural and urban environ- The role of Interactive Media Features on
ments. Journal of Environmental Psychology, the Affective Response: a Virtual Reality Study.
11, 201-230. International Journal on Human Computer
Interaction, 1(5), 1-21.
Ulrich, R. S., Simons, R. F., & Miles, M. A. (2003). Ef-
fects of environmental simulations and televi- Witmer, B., Singer, M., (1998). Measuring presence in vir-
sion on blood donor stress. Journal of tual environments: A presence questionnaire.
Architectural and Planning Research, 20(1), Presence, 7(3), 225-240.
38-38.
Zuckerman, M. (1977). Development of a situation-spe-
Valtchanov, D., Barton, K.R., & Ellard, C. Restorative ef- cific trait-state test for prediction and measure-
fects of virtual nature settings. Cy- ment of affective responses. Journal of
berpsychology, Behavior, and Social Consulting and Clinical Psychology, 45(4),
Networking, 13(5), 503-512 513-523.

JCR

View publication stats