See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228938598

Tracking restoration in natural and urban field

settings

Article in Journal of Environmental Psychology · June 2003

DOI: 10.1016/S0272-4944(02)00109-3

CITATIONS READS

716 1,265

7 authors, including:

Terry Hartig Larry Jamner

Uppsala University University of California, Irvine
125 PUBLICATIONS 5,938 CITATIONS 66 PUBLICATIONS 3,475 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Larry Jamner on 30 November 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue
are linked to publications on ResearchGate, letting you access and read them immediately.
 Journal of Environmental Psychology 23 (2003) 109–123
www.elsevier.com/locate/jep

Tracking restoration in natural and urban field settings

Terry Hartiga,*, Gary W. Evansb, Larry D. Jamnerc, Deborah S. Davisd, Tommy G.arlinge
a
Institute for Housing and Urban Research, Uppsala University, Box 785, S-801 29 Gavle,. Sweden
b
Department of Design and Environmental Analysis, Cornell University, USA
c
Department of Psychology and Social Behavior, University of California, Irvine, USA
d
School of Social Sciences, University of California, Irvine, USA
e
.
Department of Psychology, Goteborg University, Sweden
Received 26 January 2001; received in revised form 7 July 2002; accepted 30 July 2002

Abstract

We compared psychophysiological stress recovery and directed attention restoration in natural and urban field settings using
repeated measures of ambulatory blood pressure, emotion, and attention collected from 112 randomly assigned young adults. To
vary restoration needs, we had half of the subjects begin the environmental treatment directly after driving to the field site. The other
half completed attentionally demanding tasks just before the treatment. After the drive or the tasks, sitting in a room with tree views
promoted more rapid decline in diastolic blood pressure than sitting in a viewless room. Subsequently walking in a nature reserve
initially fostered blood pressure change that indicated greater stress reduction than afforded by walking in the urban surroundings.
Performance on an attentional test improved slightly from the pretest to the midpoint of the walk in the nature reserve, while it
declined in the urban setting. This opened a performance gap that persisted after the walk. Positive affect increased and anger
decreased in the nature reserve by the end of the walk; the opposite pattern emerged in the urban environment. The task
manipulation affected emotional self-reports. We discuss implications of the results for theories about restorative environments and
environmental health promotion measures.
r 2003 Elsevier Science Ltd. All rights reserved.

1. Introduction psychological demands that induced a potential for

restoration to occur. These experiments have documen-
Work pressures, urban noise, and other stressors drive ted a restorative advantage of natural environments in
many people to seek relief through outdoor recreation the ability to focus attention (Hartig, Mang, & Evans,
(Knopf, 1983; Schreyer, 1989). People are frequently 1991, Study 2) and in emotional states (e.g. Ulrich, 1979;
drawn to recreation settings such as wilderness areas Hartig, Bo. ok,
. Garvill, Olsson, & G.arling, 1996).
and urban parks by opportunities for viewing scenery, Evidence of enhanced psychophysiological recovery
contact with nature, and solitude (Knopf, 1987; Hartig, comes from a laboratory experiment in which auto-
1993). Such opportunities have been provided by nomic arousal was monitored during 10-min videotapes
planning and legislation grounded in widely held beliefs of natural vs urban environments (Ulrich et al., 1991;
that natural surroundings aid the physical and psycho- see also Parsons, Tassinary, Ulrich, Hebl, & Grossman-
logical restoration of people living in cities (e.g. Alexander, 1998; Laumann, G.arling, & Stormark,
Olmsted, 1870). 2003).
Although influential, the hypothesis of enhanced Why might natural environments better serve physio-
restoration in natural environments has only rarely logical, emotional, and attentional restoration than
faced experimental tests. In particular, few experiments urban surroundings? The experiments cited above
have compared restoration in natural and urban started from one or both of two theories. Although
environments following the controlled imposition of the two theories have some common features (Hartig &
Evans, 1993), they deal with different antecedents and
*Corresponding author. Tel.: +46-26-420-6532; fax: +46-26-420- emphasize different restoration outcomes. Attention
6501. restoration theory (ART; Kaplan & Kaplan, 1989;
E-mail address: [email protected] (T. Hartig). Kaplan, 1995) complements analyses of overload in

0272-4944/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0272-4944(02)00109-3
110 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

urban environments (Milgram, 1970; Cohen, 1978) by consistently emerged after 15–20 min (cf. Hartig et al.,
proposing factors that work in the renewal of a depleted 1996; Laumann et al., 2003), but they have appeared
capacity for directing attention. According to ART, after longer periods. Over an extended period, some of
restoration from directed attention fatigue occurs with the initial effects may dissipate. Hartig et al. (1991,
psychological distance from routine mental contents Study 2) did not find significant differences in blood
(being away) in conjunction with effortless, interest- pressure or heart rate measured after a 40-min walk in a
driven attention (fascination), sustained in coherently natural or an urban field setting (cf. Ulrich et al., 1991),
ordered environments of substantial scope (extent) when but they did find environmental effects on emotional
the person’s inclinations match the demands imposed by states and proofreading performance. Recognizing that
the environment as well as the environmental supports environmental effects on physiology might have
for intended activities (compatibility). Kaplan and emerged early in the walk and then dissipated as it
Kaplan (1989) argue that these four factors commonly came to an end, they noted a need for measuring
hold at high levels in natural environments. physiological changes during the course of a walk.
An alternative theory about restorative environments When applied to attentional performance and emotional
emphasizes the physiological and emotional changes measures as well, such a repeated measures strategy can
that can occur while viewing a scene after a situation provide insights on the relative timing of the different
involving challenge or threat. Ulrich (1983) proposes forms of restoration outcomes. That some effects appear
that perceiving particular qualities and contents in a after others have dissipated would suggest that more
scene can support psychophysiological stress recovery. than one process may have operated in producing the set
Moderate depth, moderate complexity, the presence of a of outcomes.
focal point, gross structural qualities, and natural In the experiment reported here we compared
contents such as vegetation and water can evoke positive restoration in natural and urban field settings. To track
emotions, sustain non-vigilant attention, restrict nega- restoration along different dimensions, we used multiple
tive thoughts, and so aid a return of autonomic arousal measurement methods in conjunction with a poststres-
to more moderate levels (cf. Fredrickson & Levenson, sor period that previous research suggested would
1998; Shapiro, Jamner, Goldstein, & Delfino, 2001). suffice for the environments to differentially affect
Ulrich views humans as biologically prepared to performance. With the use of ambulatory monitoring
respond positively to environmental features that signal equipment we obtained repeated measures of systolic
possibilities for survival, and so assumes an evolutionary and diastolic blood pressure (SBP; DBP) from young
basis for aesthetic and restorative responses to some adults in two successive recovery contexts: first while
natural scenes. sitting in a room with or without views of trees and then
These theories appear to complement one another while walking in a nature reserve or an area of medium-
with regard to the antecedent condition from which the density urban development. We also assessed emotional
person becomes restored. The elevated physiological states and performance before, during, and after the
arousal and negative affect characteristic of stress can walk.
occur in the absence of directed attention fatigue. The outcomes measured at a given point in time have
Conversely, elevated arousal or negative affect need to do not only with the environment available for
not always accompany attentional fatigue (Kaplan, restoration but also with the severity of attentional
1995). Yet some researchers have discussed attentional fatigue and/or stress reactions the person experienced
fatigue as an aftereffect of stress (Cohen, 1978; cf. Ulrich just before entering that environment. Thus, varying the
et al., 1991), and others have treated it as a condition antecedent condition should aid the examination of
that increases susceptibility to stress (Kaplan, 1995; cf. different forms of restoration and environmental influ-
Lepore & Evans, 1996). Thus, each of the antecedent ences upon them. Applying this reasoning, we included a
conditions may occur alone in some circumstances, but task manipulation in the present experiment with a view
in other circumstances they may have some form of to imposing intense attentional demands on half of the
reciprocal relationship or otherwise coincide. Just which subjects for an extended period just before the environ-
character the antecedent condition has determines the mental ‘‘treatment.’’
dimensions along which restoration can proceed. In sum, we experimentally tested hypotheses about
The relative timing of environmental effects along the the relative restorative values of natural and urban
given dimensions may have a bearing on whether the settings for people who had faced different kinds of
two theories address complementary processes. Differ- prior demands. In contrast to the subjects in the urban
ential effects of natural and urban environments can environment, we expected the subjects in the natural
appear quickly in physiology (within 4 min in Ulrich environment to show more rapid BP decline during the
et al., 1991; cf. Fredrickson & Levenson, 1998) and initial minutes of the treatment period (Hypothesis 1);
emotional states (within 10–15 min; e.g. Ulrich, 1979). In lower BP during the walk (Hypothesis 2); more positive
contrast, environmental effects on performance have not emotion during the walk (Hypothesis 3); more positive
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 111

change in emotion following the walk (Hypothesis 4); wildlife preserve in a canyon of the Santa Ana
and greater improvement (or a smaller decrement) mountains adjacent to Cleveland National Forest and
in performance on an attentional task following Caspers Regional Wilderness Park. Operations were run
the walk (Hypothesis 5). We also wanted to see out of plainly furnished rooms with windows through
whether environmental effects on performance would which subjects could look onto trees and vegetated
appear already during the walk, whether environmental hillsides and hear birds and a stream. The walk was
effects on physiology would persist into the postwalk along a well-graded dirt road, closed to the public, that
period, and how the varying levels of attentional runs through fields and oak-sycamore woodland in the
demands imposed prior to treatment would become canyon bottom (Fig. 1). Parking was available without
manifest in the pattern of outcomes subsequently delay adjacent to the field lab.
observed. The urban site was an area of medium-density
professional office and retail development in the city
of Orange. The area was bounded on one side by a
2. Method judicial complex and facilities of the UCI Medical
Center (UCIMC), and on the opposite side by a
2.1. Design shopping mall, restaurants, a hotel, and parking lots
(Fig. 2). The walk followed sidewalks along and across
The experimental design crossed an environmental streets of varying size, carrying traffic volumes to 24,000
treatment condition (natural, urban) with a pretreat- vehicles per day. Landscaped areas were interspersed
ment task condition (task, no-task). The environmental among buildings up to 20 stories tall. Operations were
treatment had a seated indoors phase and a walking run out of quiet, undecorated classrooms without
outdoors phase. In the natural environment, the two window views at UCIMC. Arrangements were made
phases were sitting in a room with tree views, then for parking without delay in a garage adjacent to the
walking in a nature reserve. In the urban environment, building that housed the field lab.
the two phases were sitting in a room without views,
then walking in an urban area. Note that with the 2.4. Measures
seated-indoor phase we could more readily interpret our
physiological results in relation to those from Ulrich 2.4.1. Physiology
et al.’s (1991) laboratory study. As no-task subjects We used the Accutracker II ambulatory blood
drove to the field site (a naturalistic stressor) just before pressure monitor (Suntech Medical Instruments, Ra-
the treatment, the possibility for physiological stress leigh, NC, USA) to measure SBP and DBP. The device
recovery was established in the no-task condition as well monitors the electrocardiogram via three electrodes to
as in the task condition; the task extended the duration guide auscultation through a microphone over the
of pretreatment stressor exposure and imposed acute brachial artery. White, Lund-Johansen, and Omvik
attentional demands. Subjects were randomly assigned (1990) report on validation against intra-arterial and
to groups with restrictions for equal n’s and balanced clinician measurements.
gender composition. We applied the following inflation parameters: infla-
tion to 30 mmHg above the previous systolic reading, to
2.2. Subjects a maximum of 200 mmHg, and 3 mmHg/s deflation to a
minimum of 40 mmHg. We programmed measurements
The subjects were 112 normotensive students to occur at fixed 10-min intervals.
(20.873.7 years old; 50% female; 97% non-smokers) The Accutracker appends quality control codes to
at the University of California, Irvine (UCI), screened readings in case of possible problems due to erratic or
for medications affecting cardiovascular function, missing heart beats, major arm movements, weak or too
mood, or concentration; allergies that might cause few Korotkoff sounds, or a loose cuff or air leak. We
problems in the natural environment; weight within could subsequently determine whether to exclude read-
120% of an actuarial ‘‘ideal’’ (Metropolitan Life ings on the basis of these codes and of ranges of
Foundation, 1983); and, with the women, stage in acceptable values. We accepted values from 70 to
menstrual cycle. The subjects were informed of the 200 mmHg for SBP and from 40 to 120 mmHg for
study’s nature and risks before giving written consent, DBP, given pulse pressure over 10 mmHg.
and they were compensated for participation.
2.4.2. Emotion
2.3. Environments Pretest and postwalk measures of positive affect,
attentiveness, fear arousal, sadness, and anger/aggres-
The natural environment was the Audubon Society’s sion were obtained with Zuckerman’s (1977) Inventory
Starr Ranch Sanctuary, a 4000 acre vegetation and of Personal Reactions (ZIPERS). Respondents indicate
112 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

Fig. 1. Views from within the Audubon Society’s Starr Ranch Sanctuary.

the extent to which statements describe how they feel 2.4.3. Attention
that moment (e.g. I feel elated or pleased) (1=not at all; Subjects completed the Necker Cube Pattern Control
5=very much). The ZIPERS has been a sensitive task (NCPCT) at the pretest, during the walk, and
measure in previous restorative environments studies following the walk. They first received a blank sheet
(e.g. Ulrich, 1979; Hartig et al., 1991, 1996; Ulrich et al., with a line drawing of a three-dimensional cube. They
1991). were told that their perspective on the cube would shift,
The overall happiness scale (OHS) was administered with the front and back faces of the cube reversing their
during the walk. Subjects rate their happiness on a relative positions. Once they had familiarized themselves
thermometer-like graph. Thermometer values range with this property of the Necker cube, they were
from zero, for very unhappy, to 100, for very instructed to look at the cube and tap audibly on a
happy, graded in increments of 10. Originally used hard surface when the pattern reversed. We counted the
by Campbell, Converse and Rodgers (1976) in their number of reversals that occurred during two consecu-
quality of life research, the time referent for the scale tive 30-s ‘‘hold’’ periods during which the subject was to
was changed in this experiment from a matter of days to focus on one pattern for as long as possible. Reversals
a matter of hours. The OHS has been a sensitive that occur despite the effort to hold a pattern are
measure in previous studies in this series (see Hartig thought to be due to attentional fatigue (Kaplan, 1995).
et al., 1991). We use the average number of reversals across the two
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 113

Fig. 2. Views of area adjacent to the UC Irvine Medical Center.

hold periods as the dependent variable in our analyses letters searched indicates the speed of search. The
(cf. Tennessen & Cimprich, 1995). The NCPCT has been combination of these variables (i.e. accuracy
speed)
a sensitive measure in previous studies of restorative yields an overall performance index; with the present
environments (Cimprich, 1993; Tennessen & Cimprich, version of the test, scores on this index could range from
1995). 0 to 2832. The Search and Memory test has proven
We adapted a second test of attentional performance insensitive to natural–urban comparisons in two labora-
from the memory-loaded search task used by Smith and tory experiments (Hartig et al., 1996), results that may
Miles (1987). Subjects search lines of letters for five have owed to the brief period during which the subjects
target letters given at the beginning of each line. They viewed the photographic simulations.
are to memorize the given targets, search through the
given line only once, and draw a line through any target 2.5. Procedure
found. Although they are encouraged to search quickly,
emphasis is placed on identifying all target letters. Data collection took place between late April and
Further details regarding the test materials are given by early June; the weather was typically clear and warm.
Hartig et al. (1996). The subjects performed the task for The procedure, run on weekdays, was scheduled so the
10 min. The percentage of target letters detected (% drive to the field site would occur during an uncongested
correct) indicates accuracy in the search. The number of period. As the pretreatment task took about 1 h to
114 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

Subjects were instructed to remain still during a reading

Group Pretreatment
Environmental if possible.
Treatment
After initiating the collection of baseline (i.e. refer-
Natural Stroop and Room w/ Nature
Task
Pretest Drive
Binary Tree View Walk
Post-Walk ence) BP readings, the subjects gave background
information and self-reports of positive affect, anger/
Urban Stroop and Viewless Urban
Task
Pretest Drive
Binary Room Walk
Post-Walk aggression, fear arousal, sadness, and attentiveness with
the ZIPERS. They then completed the NCPCT and
Natural Room w/ Nature
No-Task
Pretest Drive
Tree View Walk
Post-Walk SMT.
Concluding the initial phase, subjects were instructed
Urban Viewless Urban
No-Task
Pretest Drive
Room Walk
Post-Walk to drive directly to the field site, but without hurry. To
counter bias due to possible positive or negative
anticipation, the subjects were not told they were going
SBP At 10-min intervals throughout
DBP to a nature reserve or an urban setting. All received the
ZIPERS ZIPERS
same set of printed directions appropriate to the given
NCPCT EXTRA OH NCPCT field site. The routes to the sites were matched on
SMT BP NCPCT SMT
estimated travel time (40 min); potential stops due to
traffic regulations (45); and distance (23 miles to the
Fig. 3. Overview of the procedure. The top panel depicts the natural site, 21.5 miles to the urban site). Time and
progression of the different groups through the phases of the odometer mileage were recorded on departure from UCI
procedure. Task subjects began 1 h before no-task subjects, and the and on arrival at the field site, enabling a check on
task sequence (Stroop and binary classification) took about 1 h to compliance with the directions.
complete. The bottom panel relates the measurement protocol to the
phases in the procedure. SBP and DBP=systolic and diastolic blood
After the first BP reading in the field lab, half of the
pressure, ZIPERS=Zuckerman Inventory of Personal Reactions, subjects began the hour-long task sequence. Instructions
NCPCT=Necker Cube Pattern Control task, SMT=Search and and stimuli for two tasks were presented via audiotape.
Memory task. Extra BP=BP reading taken 4 min into the seated For a variant of the Stroop task, a poster was placed in
treatment phase, OH=overall happiness. front of the subject. The poster had 100 cells (20
rows
5 columns) containing color names printed in
other-colored ink (e.g. ‘‘red’’ printed in blue ink). Cell
complete, the task subjects were scheduled to begin 1 h coordinates were given every 3 s for 28 min; the subject
before the no-task subjects (12 vs 1 pm), so that both a was to say the ink color of the word in the cell. This task
task and no-task subject would be in the treatment was followed by a binary classification task; a number
sequence at about the same time in each environment.1 was presented every 2 s, for classification as even or odd
Subjects were instructed to eat a meal before participat- and high or low relative to a given criterion. This task
ing. continued for about 20 min, and terminated with a BP
An overview of the procedure is given in Fig. 3. For reading.
each subject the procedure began in a laboratory at the For task subjects the environmental treatment se-
UCI main campus. A same-sex researcher instrumented quence began immediately after the BP reading with
the subject with the ambulatory blood pressure monitor which the task terminated. For no-task subjects it began
(fixing microphone and cuff on the non-dominant arm; after the first BP reading in the field lab. The treatment
performing seated and standing calibrations; program- sequence was the same for all subjects. The first 10 min
ming inflation parameters and measurement intervals). were spent sitting quietly. Four minutes into this period
a BP reading was initiated manually. Six minutes later
the next regular BP reading occurred. The subject was
1
This meant that the no-task group spent 60 min less in the then accompanied on a 50-min walk by an assistant. At
procedure than the task group. Arguably, no-task subjects should have the outset the assistant explained that conversation
been brought into the procedure at the same time, but then made to sit should be minimized to promote consistency across
through a 60-min period without performing attentional tasks. subjects. Caution was exercised to not direct subjects’
However, such a strategy would have risked negatively affecting attention in any way. The assistants led subjects at a
emotion if the subjects were to remain inactive through the long period
(see e.g. Hartig et al., 1991, Study 2) or positively affecting emotion
slow pace (saunter) and knew where on the given route
and attention if they were given a pleasant diversion. To avoid BP readings should occur. With each reading the subject
systematic variation that would trouble the detection and interpreta- and assistant stopped walking. Twenty minutes into the
tion of effects, we allowed no-task subjects to continue with their walk (and so 30 min after the task or drive) the subjects
ordinary activities prior to participation. We assumed that attentional
completed the OHS and NCPCT following the pro-
demands and psychophysiological stressors imposed naturalistically on
no-task subjects during the hour prior to entering the experiment grammed BP reading. A few minutes after this they
would not uniformly be as intense or sustained as those imposed by the turned back toward the field lab. The procedure
task. concluded when, after returning to the field lab, the
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 115

subjects once more completed the ZIPERS, NCPCT and We checked whether drive circumstances exerted
SMT and provided additional BP measures. differential effects on BP that would cloud interpreta-
tion of treatment effects. Neither environment nor task
2.6. Statistical analysis assignment (i.e. time of departure from UCI) had a
significant main effect on mean drive DSBP or DDBP
We used analyses of variance (ANOVA) and t-tests in (based on three or more valid readings). Repeated-
the validity checks and tests for experimental effects. measures analyses were not used to check drive BP
Unless otherwise indicated, each ANOVA included effects because many subjects had one or more readings
environment, task, and gender as between-subjects invalidated (e.g. by arm movements required to operate
factors. Most of the ANOVA included time as a the vehicle). Another check was made using the last
within-subjects factor, with the number of levels drive DBP value, which would have occurred when the
corresponding to the number of measurement points. environments differed most. There were no significant
We report the Greenhouse–Geisser corrected probabil- effects of environment or task assignment on DSBP or
ities for within-subjects effects from those analyses that DDBP. Other factors—uncertainty about the destina-
encompassed more than two measures. tion, concern about time—presumably overrode the
We separately analysed the BP data for the seated- possible environmental influences.
treatment, walk, and postwalk phases. This helped us For no-task subjects the drive served as a stressor.
simplify our interpretations, conform with statistical Their mean drive SBP and DBP values were on average
assumptions, and reduce the loss of subjects from the 7.74 and 2.68 mmHg over the baseline value, paired-
repeated-measures ANOVA due to missing values. Of samples tð50Þ ¼ 4:87 (po0:001) and 2.17 (po0:05),
the 3314 BP readings provided by the 112 subjects, we respectively.
had to exclude 234 (7.1%) due to unacceptable values Performance of the task raised BP. Mean task SBP
and/or quality control codes appended by the blood and DBP values, based on at least four valid readings,
pressure monitor. Unless otherwise indicated, the were on average 4.39 and 4.27 mmHg over the baseline
analyses of BP data used change (D) scores calculated value, paired-samples tð52Þ ¼ 3:79 and 4.02, respec-
as the difference between the value for the given tively, po0:001: Repeated-measures ANOVA did not
measurement point and the baseline value obtained at find a significant main effect of environment on DSBP or
the pretest. We used the mean of at least 3 seated DDBP during the task, nor did environment interact
readings taken over at least 30 min for the baseline SBP with time in either analysis. Thus, the interpretation of
and DBP values. treatment BP effects is not threatened by differential
Degrees of freedom vary across the BP analyses for psychophysiological stress induction in the two field
the different phases due to variation in the number of settings. The within-subjects main effect of time in these
cases lost to missing values. Degrees of freedom also analyses reflects not only responses to the Stroop and
vary across analyses of the emotional state and binary classification tasks, but also response attenuation
performance data due to missing values and, in the case during each task, for DSBP, F ð5; 180Þ ¼ 18:5; and for
of the attentional measures, exclusion of subjects with DDBP, F ð5; 180Þ ¼ 5:92; both po0:001; after an initial
extreme scores (ca. 3 s.d. above the mean). We dropped peak following the onset of the task, BP declined while
three subjects from all analyses of experimental effects the task was still underway. Attenuation of the BP
due to procedural complications. response during the task made our tests for environ-
By including gender as a factor in our analyses of mental effects on subsequent restoration conservative.
variance we reduced error variance and so improved our The pretreatment tasks were meant to increase
effect estimates. However, we were not interested in attentional fatigue. Using performance data from a
gender effects per se. To simplify the presentation, we subset of the task subjects (n ¼ 44), we calculated the
only report gender effects that involve some form of percentage error for two blocks for both the Stroop and
interaction with the environment during the treatment binary classification tasks. The mean percentage error
period. Gender was not a complicating factor in the increased from the first to second block of the Stroop
validity checks that follow. task (4.29–4.98%) but declined slightly across blocks of
the binary classification task (2.95–2.89%). In an
2.7. Validity checks ANOVA with environment and gender as between-
subjects factors and task-type and block as within-
Initial two (environment)
two (task)
two (gender) subjects factors we found no significant main or
ANOVA satisfied expectations of group equivalence in interaction effects with the exception of the main effect
baseline SBP and DBP, pretest emotional states, and of task-type, F ð1; 40Þ ¼ 8:53; po0:01; the subjects
pretest performance on the NCPCT and SMT. We performed better on the binary classification task.
uncovered no significant effects involving environment Thinking the lack of a block effect might owe to the
or task assignment. very low error rate during the binary classification task,
116 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

we ran a second analysis using only the Stroop task Environment also affected BP during the walk. As
data. The ANOVA with environment and gender as shown in Fig. 4 (top panel), mean SBP shifted upward in
between-subjects factors and block as a within-subjects all of the groups between the readings at 10 and 20 min,
factor yielded a marginal effect of block, F ð1; 40Þ ¼ 3:86; reflecting the change from a seated to a standing
po0:06: No other effects were statistically significant in posture. From the reading at 20 min into the treatment
this analysis. In sum, we detected a performance decline (and so 10 min into the walk), SBP declined in the
in the initial half of the task phase, but not over the natural environment but increased in the urban envir-
course of the second and easier task. onment. Thus, the SBP trends in the two environments
We compared the drive and task as psychophysiolo- continued to diverge as they had through the end of the
gical stressors. This is reasonable; both involved a seated seated treatment. However, after the 30-min mark the
posture, vigilance, and mild physical activity (driving vs trends for all four groups converged. Those differences
verbal responding). The task subjects’ mean task DSBP seen in the pattern of change in the two environments
and DDBP did not differ significantly from the no-task underly a significant environment
time interaction in
subjects’ mean drive DSBP and DDBP. Thus, for the the analysis of the readings at 20, 30, 40, and 50 min,
procedural phase prior to treatment, mean BP was F ð3; 249Þ ¼ 2:94; po0:04: This analysis also indicated
similarly elevated in the task and no-task groups. As that, contrary to our expectations, the subjects in the
task subjects had completed the drive just before the nature reserve did not have substantially lower average
task, their pretreatment stressor exposure had longer DSBP values for the walk period as a whole; the main
duration. effect of environment did not reach statistical signifi-
A final check on the pretreatment BP status of groups cance. However, at the 30 min mark, when the trends in
analysed the final task D values and the D values first the two environments diverged the most, the mean
obtained from the no-task subjects after being seated at DSBP values differed by roughly 6 mmHg, a significant
the field lab. As the treatment began just after these difference, F ð1; 102Þ ¼ 12:97; po0:001: Thus, the results
readings concluded, they constitute the ‘‘zero time’’ D do provide some support for the hypothesis that BP
values in our analyses of change in BP during seated would be lower during the walk in the natural
treatment. The analysis uncovered no significant main environment (Hypothesis 2).
effect of environment or task assignment on DSBP or For the walk DDBP values measured at 20–50 min
DDBP, nor did environment and task interact. Thus, into the treatment, the patterns of change in the two
with respect to mean DBP levels at the onset of environments resemble those seen in SBP (Fig. 4,
treatment, the groups defined by the environment and bottom panel). Although neither the environment main
task factors were statistically equivalent. effect nor the environment
time interaction was
significant, the test of quadratic trends suggests that
environment and time interacted in a manner like that
3. Results seen in the DSBP values, (po0:04). Also, as with DSBP,
subjects in the natural environment had lower DDBP
3.1. Physiological effects values about halfway into the walk, F ð1; 102Þ ¼ 6:55;
po0:02: The task condition did not moderate the mean
While seated indoors during the first 10 min after the level of either DBP or SBP measured during the walk,
task or drive, those subjects who had views of trees nor did it affect BP change during the walk, alone or in
showed only a marginally steeper decline in SBP than interaction with environment.
those who did not have a view (po0:12) (see Fig. 4). SBP change values while seated in the field lab after
Those subjects who had just completed the drive showed the walk were similar in the natural and urban
steeper SBP declines than those who had just completed environments (see Fig. 4, top panel, 60+min). The
the task (see Fig. 4, top panel); for the task
time postwalk DDBP of subjects with tree views tended to
interaction, F ð2; 180Þ ¼ 3:63; p ¼ 0:03: differ from that of subjects seated once again in a
In contrast to the marginal effect seen in SBP and in viewless room, F ð1; 100Þ ¼ 2:84; po0:10 (see Fig. 4,
line with Hypothesis 1, the subjects with tree views bottom panel). The task condition did not affect BP
showed significantly steeper DBP declines than the measured after the walk, alone or in interaction with
subjects in a viewless room (see Fig. 4, bottom panel); environment.
for the environment
time interaction, F ð2; 180Þ ¼ 4:74; In sum, during the initial minutes of treatment DBP
p ¼ 0:01: Overall, subjects with tree views had lower declined more rapidly in those subjects who viewed trees
DDBP values during seated treatment; for the main and other vegetation in comparison to those who did
effect of environment, F ð1; 90Þ ¼ 8:94; po0:005: not have a view. Change in BP during the walk initially
Whether the subjects had just completed the pretreat- indicated a restorative advantage of being in the natural
ment task or the drive did not play a role in DBP during environment; however, the environment effect had
seated treatment. largely dissipated by the postwalk.
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 117

Fig. 4. Change in systolic (top panel) and diastolic (bottom panel) blood pressure relative to baseline as a function of environment and pretreatment
task condition. The reading at 0 min marks either the first reading in the field lab following the drive or the end of the task. The readings at 4 and
10 min occurred while subjects sat in a room with window views of trees and vegetation or in a viewless room. The readings at 20, 30, 40, and 50 min
occurred during a walk in a nature reserve or an area of medium-density urban development. The readings at 60+ min occurred while subjects again
sat in a room with window views of trees or in a viewless room.

3.2. Emotional effects and in the opposite direction, with the task subjects
reporting slightly greater happiness than the no-task
Neither environment nor task condition had a subjects (M ¼ 73:33 vs M ¼ 70:00). For the interaction,
significant main effect on OH reported during the F ð1; 99Þ ¼ 7:13; po0:01: The expectation that those
walk. However, environment and task condition walking in the nature reserve would experience
interacted. In the nature reserve, those who had more positive emotion than those walking in the
completed the task before the walk reported sub- urban environment (Hypothesis 3) receives support
stantially less happiness than their no-task coun- from the test of the simple main effect of environ-
terparts (M ¼ 67:96 vs M ¼ 79:60). In contrast, the ment within the no-task condition, F ð1; 49Þ ¼ 7:40;
difference between the two urban groups was smaller po0:01: The simple main effect of environment
118 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

among the task subjects did not reach statistical

significance.
Environment also affected pretest-to-postwalk emo-
tional change. In line with Hypothesis 4, positive affect
increased at the nature reserve and decreased in the
urban environment, F ð1; 100Þ ¼ 56:83; po0:001 (see
Fig. 5). Also, on average it increased for the no-task
subjects and decreased for the task subjects, F ð1; 100Þ ¼
9:15; po0:005: Further, environment, task, and time
interacted, F ð1; 100Þ ¼ 10:31; po0:005; the observed
means suggest that the no-task subjects showed greater
positive affect increase in the natural environment than
the task group, whereas in the urban environment the
task and no-task groups showed similar declines. Fig. 6. Change in self-reported anger and aggressiveness as a function
Also in line with Hypothesis 4, feelings of anger and of environment and task condition. Scores can range from 1 to 5.
aggressiveness declined at the nature reserve but Higher scores indicate greater anger and aggressiveness.
increased in the urban setting, F ð1; 99Þ ¼ 8:19; po0:01:
Alone, task condition did not significantly affect the
degree of change in anger and aggressiveness; however, environment. The no-task women showed similarly
task did interact with environment and time, F ð1; 99Þ ¼ small average increases in sadness in the urban and
4:97; po0:03; decline in anger and aggressiveness in the natural environments.
natural environment was concentrated in the no-task In sum, OH reported on the walk and pretest-to-
group, and increase in the urban environment was postwalk change in positive affect and anger/aggressive-
greater in the no-task group (see Fig. 6). ness were sensitive to the environment and task
Fear arousal declined slightly but not significantly manipulations. The greater OH of no-task subjects on
from the pretest (M ¼ 1:50) to the postwalk (M ¼ 1:42) the walk in the nature reserve, and the greater pretest-to-
in the sample as a whole, without showing any postwalk increase in positive affect and decline in anger/
significant effects involving environment or task. aggression, offer support for the nature restoration
Environment, task, and gender interacted in sadness hypothesis.
change, F ð1; 101Þ ¼ 6:31; po0:015: On average, men
who had completed the task became less sad in the 3.3. Attentional effects
urban environment and more sad in the natural
environment, while men who had not completed the Both self-report and performance measures indicated
task became more sad in the urban environment and less that the ability to direct attention changed over the
sad in the natural environment (see Fig. 7). In contrast course of the experiment, but they give different pictures
to the men who had completed the task, the women who of the role of the environment in that change. Self-
had completed the task on average became more sad in reported attentiveness declined substantially from the
the urban environment and less sad in the natural pretest (M ¼ 3:27) to postwalk (M ¼ 2:62), F ð1; 101Þ ¼
32:74; po0:001: However, neither environment nor task
affected the character of that change, independently or
interactively.
In contrast, environment affected change in perfor-
mance on the NCPCT. Because we wanted to know
whether environmental effects on performance had
already appeared during the walk, we first considered
change from the pretest to the walk administrations of
the NCPCT. As shown in Fig. 8, the ability to focus on
one Necker Cube pattern (and so to inhibit a reversal to
the other pattern) declined from the pretest to the walk
in the urban environment, as reflected in an increase in
the number of reversals (0.81 more reversals on average,
looking across the two task conditions). Conversely,
performance improved slightly in the natural environ-
Fig. 5. Change in self-reported positive affect as a function of ment from the pretest to the walk (0.26 fewer reversals
environment and task condition. Scores can range from 1 to 5. Higher on average, looking across the task conditions). For the
scores indicate greater positive affect. environment
time interaction, F ð1; 98Þ ¼ 13:15;
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 119

Fig. 7. Change in self-reported sadness as a function of environment, task condition, and gender. Scores can range from 1 to 5. Higher scores
indicate greater sadness.

involving environment or task condition approached

statistical significance. The number of letters searched
increased from the pretest (M ¼ 1451) to the post-test
(M ¼ 1525), F ð1; 96Þ ¼ 6:45; po0:015; here again, we
found no significant effects involving environment or
task. Despite the increase in the number of letters
searched, the slight decline in accuracy meant that
overall performance improved only marginally (po0:10)
from the pretest (M ¼ 878) to the postwalk (M ¼ 915),
again without environment or task exerting any
significant influence.
In sum, subjects reported a decline in attentiveness
Fig. 8. Change in performance on the NCPCT as a function of
environment and task condition. The values represent pattern reversals
during the experiment, an effect not modified by
that occurred despite an effort to maintain a focus on one pattern. environment or task condition. However, the natural
Across the measurement points, valid scores in this sample ranged and urban environments had contrasting effects on
from 0 to 11. change in the number of Necker Cube pattern reversals
from the pretest to the walk, opening a performance gap
that persisted through the postwalk NCPCT adminis-
po0:001: We found no significant effects involving the tration.
task condition.
Going further, we analysed change from the walk to
the postwalk administrations of the NCPCT. This 4. Discussion
analysis did not show any significant effect of environ-
ment or task, independently or interactively. Appar- 4.1. Evidence of restorative effects of natural
ently, the performance gap that had opened up near the environments
halfway point in the two walks remained largely open
into the postwalk period (see Fig. 8). So, although the First and foremost, our results speak to widely held
overall change from the pretest to the postwalk con- beliefs that natural surroundings aid the physical and
forms with Hypothesis 5, it appears that the effect owes psychological restoration of people living in cities. To
more to the performance decrements in the urban ensure a potential for restoration, we imposed different
environment than to performance increments in the demands on our subjects—tasks requiring focused
natural environment. Concluding our examination of attention, performed for an hour after arriving at a
NCPCT performance, we directly analysed pretest-to- field site, or the drive to the field site in and of itself.
postwalk change. The only noteworthy effect involving Following these demands, we found that our compar-
environment and/or task was the environment
time ison settings had opposed effects in each of the three
interaction, for which F ð1; 100Þ ¼ 5:59; p ¼ 0:02: remaining phases of the study. In the initial 10 min of
Finally, we assessed pretest-to-postwalk change in the environmental treatment, DBP declined in subjects
accuracy, speed, and overall performance (accuracy
sitting in a room with window views of trees and other
speed) on the SMT. The percentage of correctly vegetation, but it increased in subjects who sat in a room
identified target letters remained stable from the pretest without views. This result fits with our expectations
(M ¼ 62:6%) to the postwalk (M ¼ 62:1%); no effects (Hypothesis 1) and corroborates Ulrich et al.’s (1991)
120 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

findings with subjects who watched videotapes of walk; average performance in the groups did not change
natural or urban environments after exposure to a significantly from the walk to the postwalk. However,
stressor. we cannot rule out the possibility that the natural
During the next phase of the experiment the subjects environment hindered decline in directed attention
walked in a nature reserve or an area of medium-density capacity over the course of the lengthy experiment (cf.
urban development. After an upward shift in level with the self-reports of attentiveness). In contrast to NCPCT
the change from a seated to a standing posture, the performance, pretest–postwalk change in search and
blood pressure trends initially continued in the same memory task (SMT) performance did not show sig-
directions that we had seen at the close of the seated nificant effects of either the environment or task
treatment; the natural environment supported further manipulations. The SMT has previously proved insen-
blood pressure reduction, and the urban environment sitive to simulations of natural and urban environments
engendered further blood pressure increase. At 20 min in laboratory experiments (Hartig et al., 1996). The
into the walk (and so 30 min poststressor), mean DSBP present results do not help us interpret the earlier SMT
and DDBP differed significantly across the two settings. results as a matter of weak and/or too brief treatments.
This result offers some support for our expectation of Although we have converging evidence from different
lower blood pressure on the walk in the natural types of measures that the natural settings contributed
environment (Hypothesis 2). to more positive outcomes, we must emphasize that the
At 20 min into the walk we also measured OH and, as magnitude of the effects does not only owe to restorative
an index of attentional restoration, the ability to inhibit effects of the natural settings. The windowless room and
Necker Cube pattern reversals. OH scores reflected the urban surroundings had negative effects that also
joint influence of environment and antecedent condition figured in the size of the differences detected. In this
(task vs drive), and only among the subjects who had our results align with research on urban stressors (e.g.
completed the drive just before the treatment did we find Glass & Singer, 1972). Still, the changes that occurred in
the expected effect of environment on OH (Hypothesis the two natural settings had a positive character in and
3), with higher scores reported by those in the natural of themselves. The two natural settings fostered
environment. Performance on the Necker Cube task restoration; they do not merely stand as ‘‘less negative’’
improved slightly (i.e. the number of reversals declined) alternatives to the windowless room and urban sur-
in the natural environment but suffered in the urban roundings.
environment, regardless of antecedent condition. This Finally, it bears mentioning that ours was a con-
result bears on our interest in whether environmental servative test of the nature restoration hypothesis. We
effects on performance would appear already on the did not use extreme examples of natural scenic beauty
walk. and urban blight as comparison environments. Perhaps
In the last phase of the experiment, after the walk, more importantly, we did not study individuals who had
mean blood pressure D values in the two environments gone under their own initiative to some natural or urban
no longer differed as they had near the halfway point on setting expressly for unwinding, alone or with chosen
the walk. While the blood pressure effects had dis- companions. Rather, we studied individuals in the
sipated, clear environmental effects on emotion were context of a true experiment. Doing so provided validity
observed. In the natural environment, positive affect advantages, such as protection against self-selection.
had increased and anger/aggression decreased relative to However, the experimental context unavoidably im-
the pretest, while the opposite pattern of change had posed constraints on our subjects’ behavior. Yet because
occurred in the urban environment. These outcomes this feature of the study made ours a conservative test of
support our hypothesis of more positive change in the nature restoration hypothesis, we should regard the
emotion with the walk in the natural environment results obtained here as more compelling. Although
(Hypothesis 4). However, as with OH on the walk, quasi-experimental and non-experimental studies might
environment had interacted with pretreatment task report larger associations, the value of such results
condition in influencing the direction and degree of this depends on how well the researchers can address validity
change. Common to these interactions, having per- challenges such as those we have dealt with through our
formed the task in the natural environment appears to study design and procedures.
have worked against positive emotions.
Finally, postwalk performance on the NCPCT con- 4.2. Theoretical and methodological implications
firmed our hypothesis of greater improvement (or a
smaller decrement) in performance following the walk in Our results also offer some insights on two theoretical
the nature reserve (Hypothesis 5). The effect of accounts for environmental effects on restoration. To
environment on pretest–postwalk change in NCPCT improve our understanding of the relative merits of
performance appears to owe primarily to the negative attention restoration theory (Kaplan & Kaplan, 1989;
impact of the urban environment already seen on the Kaplan, 1995) and Ulrich’s (1983) stress recovery
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 121

theory, we had included two features in the experimental the physiological and attentional measures over the
design. Repeated measures of physiological, emotional, course of the experiment.
and attentional variables enabled us to examine how Other of our findings have implications for restorative
change in the different kinds of variables corresponded environments research in general. During the latter half
over the course of the experiment. Three points stand of the walk, we saw a convergence of the blood pressure
out in this regard. First, both blood pressure and Necker trends that had diverged across the two environments
Cube pattern reversals showed divergent patterns of during the first half of the walk. This pattern of change
change across the two environments through the first may have occurred because the subjects turned back
half of the treatment, increasing in the urban environ- toward the field lab at 25 min into the walk. Specula-
ment and declining in the natural environment. How- tively, this induced negative anticipation (e.g. of driving
ever, the change in Necker Cube task performance home) in natural environment subjects but some initial
correlated rather weakly with both the DSBP and DDBP relief in urban subjects. Thus, the trends may reflect
values for the corresponding time point roughly halfway shifts of emotional valence and intensity, such as
through the walk (r ¼ 0:16; p > 0:10 for both). This commonly occur during leisure episodes (Hull, Michael,
suggests that environment influenced change in these Walker, & Roggenbuck, 1996; cf. Staats, Gattersleben,
attentional and physiological variables through separate & Hartig, 1997). These results encourage caution in two
processes. respects. First, averaging multiple measures obtained in
Second, change in performance from the pretest to the comparison environments may conceal effects of those
postwalk correlated even more weakly with blood environments on patterns of change. Second, when
pressure change (r ¼ 0:12 with postwalk DSBP, r ¼ blood pressure (or some other variable) is measured
0:06 with postwalk DDBP, ps > 0:23). This seems after but not during a period in comparison environ-
unsurprising; environmental effects on performance had ments, an absence of post-test effects does not necessa-
emerged on the walk and then persisted into the rily mean that the environments did not affect the
postwalk, while the environmental effects on blood variable (cf. Hartig et al., 1991, Study 2).
pressure had largely dissipated by the postwalk. So, the Other of our results have implications for theory
environmental effects on performance again do not concerning restoration per se. These may particularly
correspond with environmental effects on autonomic interest those researchers who study the links between
arousal as reflected in blood pressure. environment, cardiovascular reactivity, and health.
Third, although pretest–postwalk change in Necker First, across the experimental phases that followed the
Cube task performance did not correlate with change in demands imposed upon the subjects, the relative effects
blood pressure over the same period, it did correlate of the environments on DBP largely paralleled those
with change in positive affect (r ¼ 0:28; p ¼ 0:004). that we observed for SBP. Yet environmental effects
The association remains significant (r ¼ 0:20; po0:05) showed up more clearly in DBP than in SBP during the
after partialling out the effects of environment; that is, seated-treatment and the postwalk phases, whereas the
we do not have a spurious association driven by the opposite held during the walk. We might interpret this
effect of environment on the respective variables. Still, pattern with reference to two cardiovascular response
we cannot say with certainty whether one type of change profiles which psychophysiological research has related
mediates the other, as we have only the two measures of to different types of stressors (Brownley, Hurwitz, &
positive affect and so cannot mount equivalent media- Schneiderman, 2000). Among other changes, increase in
tional analyses. DBP characterizes an ‘‘alpha-adrenergic’’ response
The repeated measures suggest that the physiological profile linked to stressors that involve vigilance or
and attentional restoration processes may complement passive coping. In contrast, increase in SBP helps
one another, manifesting in different kinds of outcomes distinguish a ‘‘beta-adrenergic’’ response profile linked
that emerge at different rates and persist to differing to stressors that involve active coping or defense. Thus,
degrees. In contrast, the task manipulation offers little the DBP differences during the seated-treatment and
insight into the complementarity of different restoration postwalk phases may reflect relatively less vigilance or
processes. We did not find consistent effects of the passive coping while sitting in a room with views of
manipulation on blood pressure or the attention trees, and the SBP differences during the walk may
measures. Just why the task had so little impact we reflect a less defensive orientation in the nature reserve.
cannot say. It may simply have failed to fatigue the task By implication, different cardiovascular response pro-
subjects’ ability to focus attention more so than what the files may align with different recovery contexts as they
no-task subjects faced during the hour before they began do with different stressors.
their participation and then as they drove to the field Second, on average, anger and aggressiveness declined
site. Whatever the case, we find it interesting that groups in the natural environment but increased in the urban
of subjects with seemingly different psychological points environment. Other research has shown that anger
of departure showed such similar patterns of change in impairs recovery from laboratory stressors and so may
122 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123

operate in a psychophysiological pathway linking stress Mike Kraft for photos; Peter Bowler, Bev Driver, David
and cardiovascular disease (for a review, see Linden, Ragland, and Leonard Syme for assistance at various
Earle, Gerin, & Christenfeld, 1997); however, only a stages in this project; and Staffan Hygge, Igor Knez,
tendency toward an environmental effect on DDBP Raymond Novaco, David Shapiro, Daniel Stokols, and
could be discerned in postwalk BP measures. Yet, anger an anonymous reviewer for helpful comments on earlier
reduction as a benefit of natural environments deserves drafts of this paper.
special attention, not least because anger often affects
people beyond the angry individual him- or herself,
sometimes seriously, as with interpersonal violence (cf.
Kuo & Sullivan, 2001). References

4.3. Practical implications Brownley, K. A., Hurwitz, B. E., & Schneiderman, N. (2000).
Cardiovascular psychophysiology. In J. T. Cacioppo, L. G.
Tassinary, & G. G. Berntson (Eds.), Handbook of psychophysiol-
Ineffective stress recovery may undermine physical ogy, 2nd ed. (pp. 224–264). Cambridge: Cambridge University
health through chronic arousal, immune suppression, Press.
and other aspects of allostatic load (Johnston-Brooks, Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The quality of
Lewis, Evans, & Whalen, 1998; Llabre, Spitzer, Saab, & American life. New York: Russell Sage Foundation.
Schneiderman, 2001; McEwen, 1998). An inability to Cimprich, B. (1993). Development of an intervention to restore
attention in cancer patients. Cancer Nursing, 16, 83–92.
periodically renew one’s capacity to focus may impair Cohen, S. (1978). Environmental load and the allocation of attention.
work performance and interpersonal relations. Our In A. Baum, J. E. Singer, & S. Valins (Eds.), Advances in
results illustrate how everyday settings can hinder or environmental psychology, Vol. 1. (pp. 1–29). Hillsdale, NJ:
support these different forms of restoration. As with Lawrence Erlbaum.
regular sleep, regular access to restorative environments Fredrickson, B. L., & Levenson, R. W. (1998). Positive emotions speed
recovery from the cardiovascular sequelae of negative emotions.
can interrupt processes that negatively affect health and Cognition and Emotion, 12, 191–220.
well-being in the short- and long-term. For urban Glass, D. C., & Singer, J. E. (1972). Urban stress: Experiments on noise
populations in particular, easy pedestrian and visual and social stressors. New York: Academic Press.
access to natural settings can produce preventive Hartig, T. (1993). Nature experience in transactional perspective.
benefits. Environmental strategies for health promotion Landscape and Urban Planning, 25, 17–36.
Hartig, T. (2001). Guest editor’s introduction (special issue on
that improve opportunities for restoration can offset restorative environments). Environment and Behavior, 33, 475–479.
limitations of individual-based behavioral change ap- Hartig, T., Bo. ok,
. A., Garvill, J., Olsson, T., & G.arling, T. (1996).
proaches (Schmid, Pratt, & Howze, 1995; Stokols, Environmental influences on psychological restoration. Scandina-
1996), and they complement approaches focused on vian Journal of Psychology, 37, 378–393.
preventing, eliminating, or mitigating stressor exposures Hartig, T., & Evans, G. W. (1993). Psychological foundations of
nature experience. In T. G.arling, & R. G. Golledge (Eds.), Behavior
(Hartig, 2001; King, Stokols, Talen, Brassington, & and environment: Psychological and geographical approaches (pp.
Killingsworth, 2002). Public health strategies with a 427–457). Amsterdam: North-Holland.
natural environment component may have particular Hartig, T., Mang, M., & Evans, G. W. (1991). Restorative effects of
value in this time of growing urban populations, natural environment experiences. Environment and Behavior, 23,
3–26.
exploding health care expenditures, and deteriorating
Hull, R. B., Michael, S. E., Walker, G. J., & Roggenbuck, J. W. (1996).
environmental quality. Ebb and flow of brief leisure experiences. Leisure Sciences, 18,
299–314.
Johnston-Brooks, C. H., Lewis, M. A., Evans, G. W., & Whalen, C. K.
Acknowledgements (1998). Chronic stress and illness in children: The role of allostatic
load. Psychosomatic Medicine, 60, 597–603.
Kaplan, R., & Kaplan, S. (1989). The experience of nature: A
This study was supported by USDA Forest Service psychological perspective. Cambridge: Cambridge University Press.
Cooperative Agreement 28-C8-469; an American Psy- Kaplan, S. (1995). The restorative benefits of nature: Toward an
chological Association Dissertation Research Award; integrative framework. Journal of Environmental Psychology, 15,
and a Dissertation Fellowship from the School of Social 169–182.
Ecology, University of California, Irvine (T.H.). Manu- King, A. C., Stokols, D., Talen, E., Brassington, E. S., & Killings-
worth, R. (2002). Theoretical approaches to the promotion of
script preparation was partially supported by NHLBI physical activity: Forging a transdiciplinary paradigm. American
Training Grant T32 HL07365-14 (T.H.). We thank Journal of Preventive Medicine, 23(2S), 15–25.
Susan Berg, Joanne Berry, Christine Drass, Christine Knopf, R. C. (1983). Recreational needs and behavior in natural
Fong, Anjolette Fruehan, Marcos Martinez, Dan settings. In I. Altman, & J. F. Wohlwill (Eds.), Behavior and the
natural environment (pp. 205–240). New York: Plenum.
Mathiesen, Michael Richmai, Glen Segrue, Rashida
Knopf, R. C. (1987). Human behavior, cognition, and affect in the
Tinkshell, Adrian Wolf, and Maegan Winning for natural environment. In D. Stokols, & I. Altman (Eds.), Handbook
assistance with data collection; Peter DeSimone and of environmental psychology, Vol. 1 (pp. 783–825). New York:
the National Audubon Society for Starr Ranch access; Wiley.
 T. Hartig et al. / Journal of Environmental Psychology 23 (2003) 109–123 123

Kuo, F. E., & Sullivan, W. C. (2001). Aggression and violence in the In A Literature Review: The President’s Commission on Americans
inner city: Impacts of environment via mental fatigue. Environment Outdoors (Motivations Section, pp. 1–8). Washington, DC: US
and Behavior, 33, 543–571. Government Printing Office.
Laumann, K., G.arling, T., & Stormark, K. M. (2003). Selective Shapiro, D., Jamner, L. D., Goldstein, I. B., & Delfino, R. (2001).
attention and heart rate responses to natural and urban environ- Striking a chord: Moods, blood pressure, and heart rate in
ments. Journal of Environmental Psychology, 23, 125–134. everyday life. Psychophysiology, 38, 197–204.
Lepore, S. J., & Evans, G. W. (1996). Coping with multiple stressors in Smith, A. P., & Miles, C. (1987). The combined effects of occupational
the environment. In M. Zeidner, & N. S. Endler (Eds.), Handbook health hazards: An experimental investigation of the effects of
of coping: Theory, research, applications (pp. 350–377). New York: noise, nightwork, and meals. International Archives of Occupational
Wiley. and Environmental Health, 59, 83–89.
Linden, W., Earle, T. L., Gerin, W., & Christenfeld, N. (1997). Staats, H., Gattersleben, B., & Hartig, T. (1997). Change in mood as a
Physiological stress reactivity and recovery: Conceptual siblings function of environmental design: Arousal and pleasure on a
separated at birth? Journal of Psychosomatic Research, 42(2), simulated forest hike. Journal of Environmental Psychology, 17,
117–135. 283–300.
Llabre, M. M., Spitzer, S. B., Saab, P. G., & Schneiderman, N. (2001). Stokols, D. (1996). Translating social ecological theory into guidelines
Piecewise latent growth curve modeling of systolic blood pressure for community health promotion. American Journal of Health
reactivity and recovery from the cold pressor test. Psychophysiol- Promotion, 10, 282–298.
ogy, 38, 951–960. Tennessen, C. M., & Cimprich, B. (1995). Views to nature: Effects on
McEwen, B. (1998). Protective and damaging effects of stress attention. Journal of Environmental Psychology, 15, 77–85.
mediators. New England Journal of Medicine, 338, 171–179. Ulrich, R. S. (1979). Visual landscapes and psychological well-being.
Metropolitan Life Foundation (1983). 1983 metropolitan height and Landscape Research, 4, 17–23.
weight tables. Statistical Bulletin, 64(1), 2–9. Ulrich, R. S. (1983). Aesthetic and affective response to
Milgram, S. (1970). The experience of living in cities. Science, 167, natural environment. In I. Altman, & J. F. Wohlwill (Eds.),
1461–1468. Behavior and the natural environment (pp. 85–125). New York:
Olmsted, F. L. (1870). Public parks and the enlargement of towns. Plenum Press.
Cambridge, MA: Riverside Press (reissued by Arno Press, 1970). Ulrich, R. S., Simons, R., Losito, B. D., Fiorito, E., Miles, M. A., &
Parsons, R., Tassinary, L. G., Ulrich, R. S., Hebl, M. R., & Grossman- Zelson, M. (1991). Stress recovery during exposure to natural and
Alexander, M. (1998). The view from the road: Implications for urban environments. Journal of Environmental Psychology, 11,
stress recovery and immunization. Journal of Environmental 201–230.
Psychology, 18, 113–139. White, W. B., Lund-Johansen, P., & Omvik, P. (1990). Assessment of
Schmid, T. L., Pratt, M., & Howze, E. (1995). Policy as intervention: four ambulatory blood pressure monitors and measurements by
Environmental and policy approaches to the prevention of clinicians versus intraarterial blood pressure at rest and during
cardiovascular disease. American Journal of Public Health, 85, exercise. American Journal of Cardiology, 65, 60–66.
1207–1211. Zuckerman, M. (1977). Development of a situation-specific trait-state
Schreyer, R. (1989). Motivation for participation in outdoor recreation test for the prediction and measurement of affective responses.
and barriers to that participation—a commentary on salient issues. Journal of Consulting and Clinical Psychology, 45, 513–523.