UNIVERSITY OF ' ' ILLINOIS LIBRARY AT URBANA-CHAMPAIGH
SO
NOTICE: Return or renew a.. Library Material The Minimum Fee each Lost Book is $50.00.
NOTICE: Return or renew all Library Materials! The Minimum Fee for each Lost Book is $50.00.
The person charging this material is responsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below.
Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400
UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN
FDL CATION AND SOCLU « SCIENCE LIBRARY 1 00 LIBRARY
AUU 0 7 2D|03
OCT Z I JANl 4 2J34
APR 1 4 2DM
JUN 2 7 2(06
MAR 2 3
DEC I 2
9 2009
DEC 0 8 2009
JUL
4 MA
31 arc
Z013
1 9 20U
I it *•
L161— O-1096
JUL * *
GAYLORD
UNIVERSITY OF " ILLINOIS LIBRARY AT URBANA-QHAMPAIGtl .
soc
Ar;
on or before the Laiesi uaic
Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone 'Center, 333-8400
UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN
01 1994
EDUCATION AND SODA* SCIENCE LIBRARY 100 LIBRARY
JAN 'it 6 1995
ArR \l 7 1995
AUG 161995
SEP 2^99W| 19&
NOV (
W FEB 2
MAY L 5 1996
muz
2 I99! 0
SAVLORD
UNIVERSITY OF " ILLINOIS LIBRARY AT URBANA-CHAMPAIGH
SO
MAY % 2 189b OCT 052UB2
W:
MAR 19 \M
oc
FEB24W92
^BEC13I990 ^b^2 WW30M9JJjj|^flj8|»
GAYLORD
UNIVERSITY OF " ILLINOIS LIBRARY AT URBANA-CHAMPAKM*
NOTICE: Return or renew all Library Materials! The Minimum Fee for each Lost Book is $50.00.
The person charging this material is responsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below.
Theft, mutilation, and underlining of books are reasons for discipli- nary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400
UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN
^13
HAR26
Often*.
MAY 30 1991
C L 0
(19 199?
m news
I990
0 CI
m
FEB 2^992
MM 30199
w
MAR 311992
GAYLORD
UNIVERSITY OF ' ' ILLINOIS LIBRARY URBANA-CHAMPAIGH .
JAN 12
MY 251993 JUL 16 1993
JUL 191993 JAN 041993 AUG 2H993
JAN 281993 SEP 301993
WW
THE EXPERIENCE OF NATURE
To the good earth and all things green and growing
This means acknowledging our kinship with the rest of the biosphere. If we do not feel perfectly at home here, that may after all have something to do with the way in which we have treated the place. Any home can be made uninhabitable. Our culture has too often talked in terms of con- quering nature. This is about as sensible as for a caddis worm to talk of conquering the pond that supports it, or a drunk to start fighting the bed he is lying on. Our dignity arises within nature, not against it.
Mary Midgley (1978, p. 196)
The Experience of Nature
A Psychological Perspective
RACHEL KAPLAN AND STEPHEN KAPLAN
University of Michigan
The right of 'he University of C bridge to print and sell all manner of hooks
was granted by Henry VIII in 1534. The University has printed and published continuously since 1584.
CAMBRIDGE UNIVERSITY PRESS
Cambridge
New York Port Chester Melbourne Sydney
Published by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 1RP 32 East 57th Street, New York, NY 10022, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia
© Cambridge University Press 1989
First published 1989
Printed in the United States of America
Library of Congress CataJoging-in-PubJication Data Kaplan, Rachel.
The experience of nature : a psychological perspective / Rachel Kaplan and Stephen Kaplan.
p. cm.
Bibliography: p. Includes index.
ISBN 0-521-34139-6. - ISBN 0-521-34939-2 (pbk.)
1. Nature - Psychological aspects. 2. Environmental psychology. I. Kaplan, Stephen. II. Title.
BF353.5.N37K37 1989 88-31575 155.9' 15 - dc!9 CIP
British Library Cataloging-in-Publication applied for
CONTENTS
Foreword page vii
Preface ix
Acknowledgments xi
Introduction: Nature and Human Nature 1
PART I THE PREFERENCE FOR NATURE 9
1 Perception and Categorization 13
2 The Prediction of Preference 40
3 Variations: Group Differences 72
PART II BENEFITS AND SATISFACTIONS 117
4 A Wilderness Laboratory 121
5 Nearby Nature 150
PART III TOWARD A SYNTHESIS 175
6 The Restorative Environment 177
7 The Monster at the End of the Book 201
APPENDIXES SUMMARIES OF STUDIES AND PROCEDURES
A Overview of Preference Research
Methodology 207
B Preference Studies 216
C Outdoor Challenge Program 292
D Benefits and Satisfactions Studies 300
References 318
Author Index 335
Subject Index 338
z5
V
FOREWORD
The natural environment is increasingly a source of in- terest, fascination, and affection. In a wide range of set- tings, both at home and abroad, I have found the breadth and intensity of such feelings to be remarkable. Yet until now there has been little material available for individuals who are attempting to understand their feelings toward na- ture. Though much has been written on this subject, there remains a need for a comprehensive work that is scien- tifically based, readable, and helpful.
It is clear that, were there such a volume, it would be im- portant to many people. Nature centers, arboreta, and a host of other institutions serve people's need for nature based on individual experience and intuition, but they lack a solidly grounded theoretical framework to guide their ef- forts. Many individuals are eager for a deeper understand- ing of the role nature plays in their lives. Gardeners, bird watchers, hikers, nature lovers of many varieties would welcome a way to relate intellectually to these powerful emotional forces.
The Experience of Nature is, at last, such a volume. It es- tablishes a basic understanding of nature experiences from window box to wilderness.
This foreword is addressed particularly to readers who are not trained as scientists. Although the volume is writ- ten to serve both scientists and nonscientists, the latter may benefit from a bit of guidance that the former may not require.
When I first started the manuscript I wondered whether it would be too academic to reach the wide audience that could benefit from it. As I continued reading, however, the "plot thickened," and I could see the wisdom of the ap- proach. The scientific-academic basis for the field, clearly
vii
established in part I, lays the groundwork for the rest of the book.
Although part I will be challenging for some readers, a detailed understanding of this more theoretical material is not essential to enjoying the book. And, despite the some- what technical nature of the content, the writing is clear and free of jargon; even difficult concepts are presented in a relaxed, free-flowing style.
Part II presents a more intimate view of nature experi- ences and is more easily accessible to the layperson. Here the technical concepts of perception and preference give way to a fascinating presentation of human benefits and satisfactions. These insights into human gains, central to the thesis of the book, provide much food for thought. The warm, meaningful responses presented here lead to a deeper understanding of one's self, allowing one to explore one's own nature experiences.
From part II to the end, I felt comfortable and indeed rewarded on every page, rejoicing in seeing the importance of relationships with green nature so clearly established. The book provides an important perspective for looking at one's self journeying through this world, allowing one to examine more closely what had only been suspected, tying personal nature experiences into the larger web of hu- man experience.
I think the volume creates a baseline for anyone inter- ested in the green experience. It provides the scientific basis for inquiry as well as an invitation to think more deeply about this ancient and perhaps too easily accepted role of vegetation as being "good" for people. Everyone who is interested in this relationship will find here an ap- proach that is insightful and widely applicable. They will also find a perspective that they can use as they seek to un- derstand their own nature experiences or to enhance the experiences of others. I am grateful to Rachel and Stephen Kaplan for the exciting and useful synthesis they have achieved in this important volume.
Charles A. Lewis
The Morton Arboretum
Lisle, Illinois
viii
PREFACE
Our work over the last two decades has brought us into contact with an interesting assortment of individuals - landscape architects, horticulturists, foresters, nature in- terpreters, planners, resource managers, and many more. A common denominator across their diverse fields has been an interest in the effect the natural environment has on people. These individuals have had many questions for us. Three major issues have emerged as being of particular interest. These are:
1. Is it real? Is the effect of nature on people as powerful as it in- tuitively seems to be?
2. How does it work? What lies behind the power of environ- ments that not only attract and are appreciated by people but are apparently able to restore hassled individuals to healthy and effective functioning?
3. Are some natural patterns better than others? Is there a way to design, to manage, to interpret natural environments so as to enhance these beneficial influences?
We have enjoyed working on these topics, in collabora- tion with students and colleagues in various disciplines. Gradually, over the years, the outlines of responses to these questions have begun to emerge - not definitive an- swers by any means, but ways of thinking about these issues that colleagues, students, and many practitioners say they find stimulating and useful. Gradually these ways of thinking have grown and converged. They have come to constitute both a critical mass of material and a framework that has a reasonable degree of coherence.
The time has come to think of making this material available in a more widely sharable form. The issues are not by any means wrapped up and settled. Nonetheless, there are at least two compelling reasons for taking this
ix
preface step. The first is that these many nature-related and nature-influencing activities - designing, planning, man- aging, interpreting - are going on right now, often based on little theory and even less data. Compared to the well- intended but often severely limited intuitive bases for decision making currently being employed, the results of these years of research and theory could make a positive contribution. Second, we feel the outlines of what is known and what needs to be known have become clear enough that it is time to invite others to join the fun, to par- ticipate in the process themselves.
We intend this volume to be of interest to people who may not realize that they have a common bond. Such peo- ple include gardeners, backpackers, and bird watchers. Many professionals whose work brings them in direct con- tact with the natural aspects of the environment have ex- pressed their eagerness to know about the psychological counterpart to the biological knowledge they have ac- quired. Horticulturists and landscape architects constitute two such professional groups whose training is not as rich in the psychological dimensions as many of them feel it ideally should be.
Yet another intended audience consists of our pro- fessional peers in areas such as environmental psychol- ogy, environment/behavior research, behavioral geogra- phy, environmental education, recreation behavior, and other assorted disciplines. These groups have no common professional association or single home base. We have been fortunate to have many excellent students from these areas in our classes where they have come to learn from each other and to find out that there are strong underlying themes unifying their apparently disparate directions.
Writing for such diverse audiences is a challenge. On the one hand, the topic of the book is widely appealing, and our readers will have great intuitive familiarity with the contents. On the other hand, our intention is to present results of a large number of empirical studies and consider- able theoretical material. We have tried to accommodate to these issues and to the differing expectations about sup- porting information that diverse professional groups hold. Both the chapter notes and the summaries in the Appen- dixes are intended to provide some of the information that is essential to some more professionally oriented readers but may be distracting to others. There are also references that provide fuller documentation. At the same time,
X
though our intention was to write a readable and interest- preface ing volume, we have not attempted a popularized treat- ment. Our effort is to integrate a substantial quantity of both empirical and theoretical material into a framework and perspective on the role nature plays.
From the present perspective it is difficult to believe that at the time we began this research program there were vir- tually no studies on the subject of this book. A great deal has happened since then, and the literature on the impor- tance of nature is growing. In retrospect, we find ourselves surprised by the quantity of research we and our students have done in these two decades. The present volume focuses on this more or less coherent research program. Though we refer to the work of others as it seems perti- nent, we have placed a higher priority on coherence and in- tegration than on exhaustiveness.
The task of bringing together even this comparatively unified body of research has been a most stimulating challenge. We have learned a great deal in the process. Much as we have been aware of the remarkable power of the natural environment in the lives of people, the con- vergence of the various lines of research in supporting this conclusion was even more striking than we had realized. The surprising strength of the evidence leads us to hope that those who rise to defend the natural environment against the continuing threats of encroachment will find this volume a useful ally. As an expression of support for this urgent struggle, the royalties from this book are being donated to the Nature Conservancy.
ACKNOWLEDGMENTS
We had little doubt that it would be timely to write the "Nature Book." There was substantial doubt, however, about finding the time to carry out such a major project. Without the support and encouragement of the Bloedel Foundation we would not have begun. The financial sup- port to release some time during the academic year was es- sential. Even more important, however, was the psy- chological dimension: an institution that felt our efforts needed to be shared and made the opportunity available.
In this case the institution is also a person. We first met Prentice Bloedel in 1977 through the efforts of Charles
xi
acknowledgments Lewis. Mr. Bloedel's deep-rooted understanding of the im- portance of nature expressed itself in many ways. The Bloedel Reserve on Bainbridge Island, Washington, is one such expression. We had the good fortune to walk across this land with him as our guide. His brief, modest com- ments were invariably perceptive. His interest in our work and support of this book are yet a further expression. Our thanks to him are many.
We are also grateful to the U.S. Forest Service for con- tinuous support of our research since 1970. The Forest Service gets its share of criticism for some of the most vis- ible aspects of its work. The support we have enjoyed, though far less visible and on a much smaller scale, has nonetheless amounted to a substantial contribution. Walt Hopkins and Duane Lloyd in the Washington office in the 1970s, Jim Morgan and various successors in the St. Paul office, and particularly John Dwyer, project leader of the Urban Forestry Project for the North Central Forest Ex- periment Station in Chicago, have given us a different view of the Forest Service. Their support has extended far beyond the financial; they have actually been interested in the substance of our work and have been excellent dis- seminators of some of its results.
This book represents the efforts of many. The numerous studies featured in the appendixes include the works of former students and colleagues who have shared our in- terest in the natural environment. Many of these in- dividuals have contributed to the development of our ideas. Many others have also left a strong mark through their critical comments and searching questions.
The book itself benefited from the efforts of several in- dividuals. Three in particular made substantial contribu- tions: Thomas Herzog commented on several chapters; Lisa Bardwell took major responsibility for preparing Ap- pendix B; Janet Talbot, a colleague of long standing, pre- pared Appendix D and collaborated on Chapter 5.
Our gratitude and indebtedness are far greater than this listing would suggest. We have had the good fortune to work with outstanding, dedicated students, to be constant- ly barraged by eager, thoughtful individuals who have run across our work, and to have been part of a community of seekers who share our conviction that the natural environ- ment plays an important role in human well-being.
xii
INTRODUCTION: NATURE AND HUMAN NATURE
Nature is a valued and appreciated part of life. Ex- amples abound. People plant flowers and shrubs and nurture house plants; cities invest heavily in trees; citizens band together to preserve natural settings they have never seen; landscapes for centuries have been the subject of painting and poetry. Nature seems to be im- portant to people. Though substantial sums of money are spent on nature and natural settings, it is hard to justify the role nature plays in rational terms. In fact, people with relatively little money are no less likely than the more af- fluent to have a splash of colorful flowers in front of their homes. Bond proposals for parks have often passed even when other issues fail. The grief neighbors feel when "their" tree is removed can hardly be explained on eco- nomic grounds.
It is no doubt possible to provide alternative ex- planations for any one of these examples. As a group, however, they provide at least circumstantial evidence that nature is important in itself rather than for some ex- trinsic reason. Further evidence in support of this hypo- thesis is provided by several recent studies using special populations. Verderber (1986) has shown that the quality of the view out the window is a significant factor in the recovery of patients in physical medicine and rehabilita- tion wards of six hospitals. Ulrich (1984) demonstrated that the content of the view is important in hospital patients' recovery from surgery, with nature content con- tributing to faster recovery. Moore's (1981) study showed a dramatic relationship between inmates' use of health care facilities at a large federal prison and the view from their cell. Those whose view was of other inmates sought health care most often. Of the inmates whose views were of areas outside the prison building, the ones who looked out onto
l
introduction the surrounding farmland sought health care least of all.
West's (1986) study further supported these findings in another prison setting.
These carefully designed studies established clear sup- port that nature in itself plays an important role. After all, it would be most unlikely that the patients and inmates were "using" nature for some other goal, such as to achieve status, to gain personal identity, or to establish territory. The purpose of each of these studies, however, was not to explain why nature plays such an important role, but to make it clear that there is a meaningful phenomenon to examine and that there are important im- plications in terms of basic human needs.
Many of the studies reported in this volume provide further evidence for the importance of nature to people, but that is not our primary purpose. Rather, our intention is to go beyond this, to focus on what nature does, for whom, under what circumstances.
ABOUT NATURE
This book is about the natural environment, about people, and about the relationship between them. It is about things many have known but few have tried to study empirically. It is about things for which there is only a limited vocab- ulary.
Nature connotes many settings. As is clear already, our use of the word nature is intended to be broad and in- clusive (although, for the most part, the discussion ex- cludes fauna). The discussion of nature here is not limited to those faraway, vast, and pristine places where there has been little human intervention, or to places designated as "natural areas" by some governmental authority. Nature includes parks and open spaces, meadows and abandoned fields, street trees and backyard gardens. We are referring to places near and far, common and unusual, managed and unkempt, big, small, and in-between, where plants grow by human design or even despite it. We are referring to areas that would often be described as green, but they are also natural when the green is replaced by white or brown or red and yellow.
Nature includes plants and various forms of vegetation. It also includes settings or landscapes or places with plants. Thus the plants and their arrangement in a space,
2
and the juxtaposition of the plants to other elements in the environment, all play a role in our discussion.
The expression natural environment is not intended to include only purely natural elements, any more than the buiJt environment refers exclusively to constructed ele- ments. Similarly, the contrast between natural on the one hand and urban and rural on the other we find to be un- helpful. Much of our discussion is about the nature that can be found in the urban and in the rural context.
It is clear that whereas the concept of nature is very much part of the human experience the language for dis- cussing it is neither rich nor precise. Although nature and natural environment as used here are difficult to define adequately, they refer to things and places we have all experienced.
ABOUT PEOPLE
The studies discussed in parts I and II explore different natural settings and diverse facets of the way people relate to such places. Though these studies have been guided by a theoretical perspective, it is even more clearly the case that the theory has been guided by the studies. Each of the chapters deals with both the studies and the explanatory framework. It may be helpful to provide an introduction to this framework. It is a view that considers humans as deeply concerned with information and examines the en- vironment as a vital source of information.
Human functioning depends on information.1 Much of this information is provided by the immediate environ- ment. There are signs, both verbal (such as a street name) and nonverbal (such as a doorbell), that provide guidance to behavior. There are combinations and arrangements of elements that constantly require deciphering: a group of people standing near eath other, all facing the same way; a long hallway with many doors; a series of cash registers; a group of trees where two paths cross. Some of the informa- tion is urgent and requires action; some is made urgent by size, motion, or color and may be difficult to ignore despite being irrelevant to one's current goals.
A great deal of the information that is essential to functioning is already stored in our heads based on pre- vious experience. Such stored information not only makes it possible to assess a current situation but is essential in
NATURE AND HUMAN NATURE
3
introduction anticipating what might happen next. Humans can close their eyes and imagine, they can consider alternate plans, they can give advice or instructions - all based on informa- tion that is not immediately present in the environment.
It is clear from this discussion that information is not only what scholars consume and intellectuals exchange. In fact, it is reasonable to say that information (or knowledge) is the common human coin. The storage and processing of information are the cornerstone of human functioning. In addition to relying on the information that has already been acquired, humans are active in their pursuit of ad- ditional information. In fact, they often seek information even when having it makes little discernible difference. In- deed, the interest in useless information is so powerful that, according to Postman (1985) in his provocative Amus- ing ourselves to death, it is used against us by the media. He considers television, newspapers, and Trivial Pursuit to be striking examples of this unfortunate turn of events.
People are particularly aware of information that is visual, that concerns what they see. That does not mean that people interpret the information in visual terms ex- clusively; rather, visual stimuli are effective in conjuring associated information. The sight of water provides inf or- • mation about potential opportunities^ which may or may not be visual in themselves. Magazine advertisements rely upon the reader's imagination (and prior experience) to recognize other aspects of the scene than the visual materi- al that is presented. Visual material is thus particularly ef- fective in evoking other kinds of information that had pre- viously been associated with the presented information. (The expression I see is an interesting manifestation of the dominance of the visual mode without the necessity that it refer exclusively to visual information.)
Similarly, humans are strongly oriented to spatial infor- mation. A nearby moving object (for example, a person passing one's office door) is difficult to ignore. A great deal of information that is not necessarily spatial is, in fact, coded in spatial terms. There are many examples of this in the language (e.g., reference to higher orders, bottom line, feeling high or low or under it). A great deal of problem solving and thinking involves arraying pieces of the prob- lem in a real or imaginary space. Thus, both conceptually and with respect to information in the current environ- ment, people are sensitive to relative location.
One m^nTitelnTs^^ in our brief over-
4
view of the theoretical perspective that guides this book. nature and We have said that information is essential to human human nature functioning, that people store and use huge amounts of in- formation, and that they actively seek more information. None of this speaks to the strong feelings people have about information. The relationship we humans have to in- formation is, in many situations, far from neutral. We assess current and future situations in terms of whether they are/will be good or bad, pleasant or painful. The an- ticipation of an unpleasant situation, even if one is cur- rently in a pain-free situation, can make one's feelings negative. Similarly, people whose conditions are ex- tremely painful can feel much better if they anticipate a hopeful future.
From this perspective, then, people not only are adroit in their use of information but crave it and continuously evaluate it. The evaluation of information is essential; ef- fective functioning relies on sorting the goodfrmrilhelDad, the useful from the useless. Humans judge situations with such facility that they are often not aware of the fact that such an evaluation is occurring. The rapid intuitive evalua- tion by people of other people is a widely recognized phenomenon; the evaluation of environments occurs with comparable unobtrusiveness and comparable facility.2
ORIENTATION TO THE BOOK
This book is a reflection of an odyssey. We have been in- terested in the relationship between people and the natural environment for nearly 20 years. During that time the varied interests of students we worked with and the diver- sity of research opportunities have left their mark. As a result we have become involved in a rather wide spectrum of topics in this area, including scenic assessment, garden- ing, wilderness recreation, and what we have come to call "nearby nature." At many points along the way these various studies did not seem particularly related, but gradually, from these separate parts, a coherent sense of the whole has begun to emerge. As a reflection of this con- tinuing odyssey, this volume is based largely on our own work and that of our students (although the work of others is, of course, referred to as well). Even with respect to our work, this volume is an attempt not to be exhaustive but to
5
introduction emphasize the larger conceptual themes and to bring in # jpecific_ studies as a means of makingjhese them^sjnore concrete and understandable.
There are three major sections to the book. The last of these, part III, provides a synthesis of what came before. In it we develop the concept of a restorative environment - an environment in which the recovery of mental energies and effectiveness is enhanced. The examination of factors that make it restorative draws on the issues of preferences, satisfactions, and fascinations, the foci of the first two sections.
The first major section of the book, part I, is devoted to research on the perception of and preference for natural settings. Given a species that is strongly oriented to visual and spatial information, and is quick to have feelings about things, what role does nature play? Are there natural elements and natural settings to which people react strong- ly, and do such reactions show consistent patterns? A par- ticularly important aspect of this section is an understand- ing of the role that preference plays. Rather than being frivolous, preference as we have come to see it through numerous studies is an expression of a deep and underly- ing aspect of human functioning. Although people are generally not aware of it, their preferences are influenced by such factors as whether they could learn more in a given environment and whether they would be able to move around in the environment without fear of getting lost.
Consistency in the pattern of preference for natural settings does not, however, suggest universality in what people prefer. Part I also explores variations on the theme - different ways in which such preferences express themselves. Although in general there are broad areas of agreement as to what people like in the natural environ- ment, there are at the same time significant and consistent differences. These are interesting both for what they add to the understanding of preference in general and for their implications for the design and management of natural settings.
Part II deals with research on the satisfactions and benefits people derive from contact with natural environ- ments. Research we have carried out in a wilderness con- text over a period of 10 years provided a laboratory for studying such satisfactions. Fascination appears as an im- portant component. Compatibility, the match between the demands of the environment and the goals and intentions
8
of the individual, is a particularly strong factor in the nature and wilderness context. human nature
Similar kinds of satisfactions and benefits to those derived in the wilderness are also evident in the immediate natural setting. The nearby natural environment is vital both to how satisfied people are with where they live and to their very quality of life. Research on the satisfactions people obtain from gardening points to the central role of fascination here, too. In the presence of such fascination people are able to rest their minds in a way that allows them to recover their effectiveness.3
NOTES
1. Although the concept of information is central to our ap- proach, our version of the information-processing position is relatively unconventional. It views the processing of infor- * matijnjn_biological systernTas differerlMiTm^Ti^rrespects from^the way compxUe^s^r^essintbnnaTio also takes several steps in the directiorToTGihson (1979), an outspoken critic of the information-processing approach. Nonetheless, compared to most other approaches within environmental psychology, our emphasis on the role of information and the importance of tracing its flow thro^ighthec^ajnsjn places us
at least on the_fringe_ofthe im^mation-p^o^e^sing approach. For an exposition of this perspective, see Cognition and en- vironment: Functioning in an uncertain world (S. Kaplan & Kaplan, 1982).
2. Implicit in this discussion is the assumption that much of human functioning is impacted by its evolutionary origins. We have chosen not to emphasize this relationship in this volume since the theory and data presented here speak for themselves and are not dependent on this evolutionary back- ground. Readers interested in exploring this potentially en- riching theoretical linkage may wish to look at the pertinent portions of Humanscape and cognition and environment (S. Kaplan & Kaplan, 1978, 1982).
3. In addition to these three major sections the book has four ap- » pendixes. These include summaries of a substantial number
of studies that are the grist for the material in parts I and II. Rather than interrupt the flow of discussion, weliave marked these studies with an asterisk when they are mentioned in the text, which indicates that they are described in the appen- dixes. Although solne~reaQ^rswiir find little reason to turn to any of this material, others will find it useful for more detailed study of the empirical background behind the ideas presented here.
7
PART I
THE PREFERENCE FOR NATURE
here is the story of the fellow who is sitting at his
rather bare desk when the phone rings. He picks up
JL the receiver and responds to the caller's query, "Yes, we do landscapes." Some seconds later: "Well, whatever you like. How about one with, say, some mountains . . . and a river running through there, and a canoe paddling down the river?" And then, "You don't want a canoe? Oh, that's no problem. Sure. We can do it without. Yeah. OK. We can have that ready for you in a week." As soon as he puts the receiver down, the fellow is heard calling to the back room, "Hey Joe, one landscape and hold the canoe."
Though the canoe may be optional, the presence of water is highly likely in a made-to-order preferred land- scape. It can be an ocean, a big lake, a small lake, river, stream, or pond; it might be placid or fast-moving, tranquil or falling, with trees reflected or with rapids. Water is a highly prized element in the landscape.
Water is not just an attractive element in pictures. Peo- ple are willing to pay higher rents for a water view. In fact, along many bodies of water the density of residential development provides an all too vivid indication of peo- ple's willingness to be near the water even if they must share the view, the facilities, the roads, the traffic, and everything else. The fondness for the water seems to hold whether it is a place for active water sports or not, whether one plans to be "using" the water or is unlikely to ever directly interact with it. It turns out that much of the enjoy- ment that waterscapes provide is low in action.
Water provides an excellent example of an aspect of the natural environment that is highly preferred. Though water seems to be an attractive element, it is also the relationship of the water to its surroundings that is impor- tant in the preference. It is, thus, meaningful to examine
9
not only the specific "things" or elements or contents of the natural environment that are preferred but also the relationship among these elements. A landscape is more than the enumeration of the things in the scene. A land- scape also entails an organization of these components. Both the contents and the organizational patterns play an important role in people's preference for natural settings.
Before we launch into an examination of preference, it is important to be clear about why this is a worthwhile topic at all. Preference has a frivolous connotation. It suggests the decorative rather than the essential, the favored as op- posed to the necessary. It also seems idiosyncratic - tastes are known to vary, after all. Given that the world is less than perfect, that many people do not even have their basic needs met, preference may appear to be a luxury that only few can afford to consider.
The view we take of environmental preference is in strong contrast to such a position. Preference is intimately tied to basic concerns. We see preference as an expression of underlying human needs. Preference can be expected to be greater for settings in which an organism is likely to thrive and diminished for those in which it may be harmed or rendered ineffective. Thus humans, like other animals, are far more likely to prefer a setting in which they can function effectively.
Aesthetic reactions thus reflect neither a casual nor a trivial aspect of the human makeup. Rather, they appear to constitute a guide to human behavior that is both ancient and far-reaching. Underlying such reactions is an assess- ment of the environment in terms of its compatibility with human needs and purposes. Thus aesthetic reaction is an indication of an environment where effective human func- tioning is more likely to occur.
Such a position does not require that people are neces- sarily aware of their needs or that preferences be univer- sal. The way preference feels to the perceiver stands in sharp contrast to the process that underlies it. Preference feels direct, immediate, and holistic. One experiences no hint that one is going through a complex, analytic process en route to one's judgment. At the same time, research results point to a variety of variables that must be assessed and integrated in making a preference judgment. Such an underlying process must be both very fast and quite unconscious.
Part I consists of three chapters. The way the environ-
10
ment is perceived or experienced is the focus of the first of THE preference these. Perception and preference are closely related. Per- F0R nature ception is a key element in preference, and the measure- ment of preference permits an examination of the percep- tual process. Perception is also strongly influenced by previous experience. The same environment may be per- ceived in distinctly different ways by a planner, a de- veloper, and a neighbor. The emphasis in chapter 1 is on the way the natural environment is experienced by people with no particular training in environmental domains. Even though each of us has had extensive experience in such settings, our awareness of the underlying categories of the experience is at best minimal. Through studies that focus on preference, however, it has been possible to ex- tract what these categories are. These underlying cate- gories of the way one experiences the environment pro- vide considerable insight into what it means to "function effectively."
In chapter 2 the focus is on factors that seem to account for preference of natural settings. These factors include both contents - particular kinds of settings and natural ele- ments (such as water) - and patterns of organization of the natural setting. As it turns out, these factors do not con- stitute a long list. Rather, they can be understood easily in terms of the human inclination to seek and process infor- mation. Thus the relationship between the preference fac- tors and the informational patterns plays a central role in the chapter.
Chapter 3 deals with some variations in the ways these major informational themes and preferences express themselves. Cultural and ethnic patterns can be reflected in distinctly different preferences for kinds of natural set- tings. There are also other sources of variation in prefer- ence based on people's experiences. For example, in- dividuals whose experience includes professional training in design-related fields may show distinctly different pref- erence patterns. Variations are important to identify and understand if one takes the notion of preference seriously. Understanding both the common themes and their varia- tions is essential if environmental decision making is to be responsive and humane.
11
PERCEPTION AND CATEGORIZATION
ppraisals have traditionally been the province of
experts. There are experts who evaluate jewels,
A* JL antiques, homes, paintings, and many other ob- jects. There are also experts who assess the aesthetics of the scenery. What all these individuals have in common are training and experience that enable them to hold a standard against which they compare a particular example that is to be assessed for its value. One gold watch might be worth $300 whereas another is appraised at 10 times as much. A particular section of highway may be designated as scenic whereas other sections fail to achieve such designation. Such evaluations are considered to have ob- jectivity both because they are judged relative to standards and because the appraiser is assumed to make these judgments in terms of established standards rather than personal reactions.
The process of making these appraisals requires that one categorize or classify. To decide that a particular painting is worth a certain amount depends on classifying it in terms of various designations: artist, age, condition, frame, and so forth. To make decisions about the permissibility of changing a setting - for example, in siting a building in a national park - the current setting must be classified in terms of such considerations as its present use, the sen- sitivity of the area, its uniqueness. Thus, in achieving ex- pert judgments as to the value of an object or setting, it is ^necessary to make a series of decisions about the cate- gories to which it belongs.
The choice of appropriate categories for making an ap- praisal depends on the purpose of the assessment, but it is also affected by the expert's area of training. The ap- praised value of the same object may differ depending on the specialized knowledge of the appraiser. The nature of
13
the appraisal may also differ if the purpose is to set a purchase price or to provide a plan for modifying it. Thus professional advice about changing the grounds around a corporate headquarters may call on factors other than the monetary value of the site but may still require a series of categorizations about the value of what is there now, the kind of setting, who will use it, and so on.
It has been traditional to make a distinction between expert-based appraisals and the preferences that people have for the same objects or places. people's preferences are not guided by special training, nor do they have knowl- edge about the appropriate standards. For such reasons, preferences - as opposed to appraisals - are thought to be discrepant and subjective. Nonetheless, there are strong similarities between the process used by the expert in mak- ing an appraisal and by the untrained individual who is making a preference judgment. For both expert and nonex- pert, a great deal of experience is brought to bear on the decision, and categorization is necessarily an aspect of the decision. Neither expert nor nonexpert is likely to be aware of the process or categories that lead to the decision.
This chapter is about the categories that are implicit in the experience of the natural environment for individuals who have no particular training in environmental fields. Although geologists, for example, may be cognizant of the essential categories underlying their work, the underlying categories for the untrained person are unlikely to include bedrock and moraines. But what are the categories? Given the lack of training in appraising the environment, one could expect the categories to be idiosyncratic and as dif- ferent as people are different. This has not proven to be the case. Rather, we have found strong commonalities in these categorizations. Because people are rarely aware of how they experience an environment, identifying these com- mon themes in the perceptual process requires an indirect approach. As it turns out, preferences provide a means for discovering the categories of perception. Let us begin, then, by describing the research approach that made it pos- sible to study these categories.
THE RESEARCH FRAMEWORK
Back in the time of Earth Day there was growing concern about environmental degradation and increasing con-
14
sciousness about the ugliness of the physical surround- perception ani ings. It became apparent to us (trained as experimental categorizatio psychologists) that in very important respects the environ- ment was ignored in a great deal of psychological research. Research on aesthetics, potentially a pertinent topic, was largely based on how much people liked artificially con- structed stimuli - patterns designed to differ in their amount of complexity. Such material has great advantages to the scientist. The issue of expert appraisal could be totally ignored as few people would have extensive experi- ence with such material. Furthermore, by constructing the stimuli the researcher has full control over their properties. And control is, after all, of utmost importance to the scientist.
Knowledge gained about preference for such patterns may have little relationship to what people like and dislike in the world about them. To find out about preference for existing settings, however, one must give up some of the control. Since the purpose of scientific control is to in- crease confidence that one is, in fact, isolating the effect of particular factors, reduction of such control must also reduce one's confidence. This is an unavoidable trade-off; one cannot have it both ways. The studies we will be dis- cussing flow from the judgment that control over irrele- vant material is of less utility than knowledge gained about relevant material, though under less than optimal con- ditions. The way to gain confidence despite this limitation is by the accumulation of knowledge across numerous studies. If the pattern of results turns out to be consistent despite great diversity in both the settings that are studied and the backgrounds of the study participants, one's con- fidence is considerably bolstered.
The research program that we started at that time (around 1970) thus focused on scenes of the environment rather than on artificial stimuli. In other respects, how- ever, the research followed previous work in its basic ap- proach. Individuals were asked to look at each stimulus - now a scene - and to indicate how much they liked it. In our early studies the scenes were presented for a fixed time, but later studies - using printed photographs rather than slides - permitted individuals to pace themselves. As was true of earlier psychological research, we also had the scenes rated in terms of other qualities, which could then be related to preference.
Such a procedure permits use of meaningful material
15
the preference while still maintaining some elements of control. The re- for nature searcher selects what is to be viewed. The scenes are pre-
sented as slides or photographs, thus eliminating some of the experimental "noise" of the actual physical setting. A fairly large number of scenes can be presented using this method. This makes it possible to examine more carefully the factors that affect the preferences,
Once again, as soon as such controls are introduced the trade-off dilemma arises. Would it not be better, or more accurate, to take people to particular settings and have them rate these for preference? After all, photography can be deceiving. One can take pictures so as to make a setting more pleasant than it actually is. The "noise" in the real setting brings in other sense modalities - sound, smell, touch. Are judgments made in response to slides of settings in any sense comparable to those that would be made in the setting itself?
The quick and straightforward answer to these ques- tions is that people's responses to the two-dimensional representation are surprisingly similar to what they are in the setting itself. A number of studies have focused on these issues because these are basic questions that needed to be resolved.1 Though the similarity in responses to pic- tures and to real settings may seem surprising to those who focus on research methods, it is much less surprising if considered in terms of daily human experience. Much of the information that we consider all the time reaches us by means of two-dimensional representations of three-dimen- sional settings. When watching television or seeing pic- tures in a book or a painting on the wall people are not likely to say that the representation is deceiving.
It would, in fact, be much more difficult to justify a pro- cedure that relied upon preference judgments made in the physical environment itself. The process of bringing the study participants to each of the various settings would in itself introduce a great variety of problems. It would also be difficult to ask the participants to consider as many set- tings under these circumstances as can easily be used with photographs.
Before we leave these methodological concerns, there is one more that needs to be mentioned. Although our earliest studies used color slides to represent the settings, many of the later studies relied upon small black-and-white printed pictures. This raises the question of people's ability to judge material that is not in color and where the photo-
16
graphic quality is short of outstanding. Again, the quick perception and answer is that this poses no problems. Participants might categorization express regret that the scene is not in color, but their ability to react to it is not hampered. Participants also often "read" color into the scene. Natural settings, for instance, may be seen as green despite the absence of color. Though there certainly are some advantages to color, it has disad- vantages as well. Color that is poorly matched to one's ex- pectations (lawns that are too blue, for example) creates greater difficulties than black and white. Furthermore, the use of black-and-white photographs makes it possible to distribute the material in printed form at minimal cost.
The material presented in this chapter is based on research that used photographs of real environments and asked people to judge these for preference. In all cases in- dividuals were asked to rate each scene in the study using a 5-point rating scale (where 1 means they "don't like it" and 5 means they "like it a great deal"). Each of the studies in- volved various other aspects and focused on a different kind of setting and on a variety of other issues.
A substantial number of studies that have been carried out using this procedure are not discussed in detail within the chapter. Rather, many of them are presented in sum- mary form in Appendix B (and are noted by * when men- tioned in the chapter) so that the treatment here can focus on the insights gained from the studies as a whole. Neither the discussion nor the appendix, however, is exhaustive. By now there are more studies that accrue to this line of research than we even know about, and there are others that are sufficiently redundant with what is included that they did not seem needed.
CATEGORIZATION
Professional training involves learning to see things in a particular way. Combinations of elements that had pre- viously not been associated with each other for the novice are given a name that others, trained in the same area, also recognize as salient. The elements might be symptoms that in combination can be recognized as a disease. They may be treatment of doors, windows, and roof lines that com- bine to be designated as particular architectural styles. Such elements might also be different patterns of cultiva- tion or of land use or of many other things. This learned
17
the preference way of seeing particular combinations of elements is es- for nature sential to the experts who share it. To call something a
mixed stand or a storm drain or a subdivision achieves a shared understanding that carries a great deal of informa- tion that can remain unstated because it is implicit in the name. During the brief period when one is acquiring such new information and learning to identify the combination 9 of elements in terms of its new appellation, one may be sensitive to the fact that a new way of seeing is indeed developing. Before long, however, the new terminology is a part of one's vocabulary, and what had been a new way of seeing seems no different than the way one had seen things all along.
When trying to explain some of these newly learned concepts to someone else, one often realizes that the con- cept has become complex and that the explanation is far from straightforward. Much more than a vocabulary has been acquired. A process of perceptual learning has taken place that includes distinctions that may be difficult to ver- balize. Having seen various instances of the new concept leads to a richness of meaning that goes beyond any single example. This very richness had presented difficulties earlier in learning the concept, and now, in trying to com- municate the concept to someone who is new to it, the mul- tiplicity of its facets becomes evident.
None of this is very different from the rest of our knowl- edge and experience.2 The things we know about, even in areas where we do not consider ourselves to be expert, follow the same patterns. We see things in particular ways and have names for certain combinations of elements; these names apply to a richness of patterns that makes definition difficult. All of us use words like park and gar- den and nature. We have a shared understanding about them. What constitutes a park? Do parks have swings? benches? walks? No simple definition suffices . We can ac- cept the notion that parks differ and yet they are parks (although it may be more difficult to accept "research parks" and "industrial parks" as included in the same con- cept). A backyard garden may be like a park in many re- spects, but we would usually not include those in what we designate a park.
These issues are of utmost importance in considering perception. Though there are categorical distinctions un- derlying perception, we generally do not realize what these categories might be. Preference judgments, in turn, are
18
based on perceptions. The research on preference thus perception and tries to determine not only what people do and do not like categorization but also what some of the categories are that constitute the basic patterns of daily experience.
Category-Identifying Methodology
The categorizations that experts use in particular settings are often not meaningful (and may even be disturbing) to those who do not share that expertise. The nonexpert "doesn't see it that way." For example, uses that a planner may categorize as "institutional" a citizen may see as hav- ing little in common, as their impact on a neighborhood may be radically different. The citizen is thus categorizing uses in other terms than the planner is. An administrator responsible for various penal institutions may have a quite different way to categorize institutions than either the planner or the citizen. There are thus numerous ways to classify different kinds of environments; each may be cor- rect and appropriate in its context, but they may clash when the contexts differ. Though the categorizations of many areas of expertise are often identifiable, those of common perception - of the citizen, as it were - are less self-evident.
Categorization is just as central to the natural environ- ment as to any other area of experience. The management of huge tracts of public lands depends on classification. Both the U.S. Forest Service (1974) and the U.S. Bureau of Land Management (1980) use such classification systems in their management of the public's visual resources. The "character types," "variety classes," and land-form and land-use categories central to these systems do not corre- spond to the way an untrained citizen would categorize or "see" the environment. Decisions about major changes to the natural environment are based on these expert cate- gorizations.
For each area of expertise, such categorizations take on an air of reality; for the group sharing the code, the categories appear to be reasonable, widely accepted bases for describing the respective environment. It is in the na- ture of expertise that one not only perceives the world through these categories but no longer remembers that it was ever otherwise. But the basis for environmental cate- gorization for the person-on-the-street, as it were, may not
19
match the land-use, land-cover, forest-practice, or biophy- sical distinctions.
People react to what they experience in terms of com- monalities, in terms of classes or categories. A scene is generally perceived as a particular instance of a larger class of scenes. Let us say we ask people to categorize 40 scenes showing different outdoor environments into groupings they consider important. A park planner may categorize them in terms of activities and their impacts, a botanist in terms of species. The groupings of other in- dividuals may reflect distinctly different aspects of the set- tings. What might these be?
By examining the preference judgments from a par- ticular study, one can determine some of these com- monalities. The procedures for extracting this information - referred to here as Category-Identifying Methodology (CIM) - are discussed in Appendix A.3 For purposes of the discussion here, it is important to realize that this approach depends on several issues. Clearly, research of this kind re- quires that there are enough individuals participating in the study. The interest, after all, is in how "people" see the environment and what they prefer. Perhaps less clear is the requirement that one provide enough instances of the kinds of environments one is studying. The issues in sam- pling the environment are no different than those involved in sampling people. In both cases, one is concerned to determine relatively stable findings that go beyond the par- ticular instance. If we know the preferences of a handful of people, we cannot be confident that these are similar to the preferences of others. If we know preferences for one or two examples of a particular kind of setting, we cannot be confident that these are similar to preferences for other instances of the same category.
Preference ratings thus generate diverse information. One can examine the ratings in terms of how much various scenes are liked or disliked by computing the average rat- ing for each scene. (Results based on such analyses will be discussed in chapter 2.) One can also examine the pattern of the ratings, their relationship to each other. Analyses based on such patterns provide insight into shared cate- gories of perception. Using Category-Identifying Method- ology, one can thus extract information about how scenes are grouped. The scenes constituting a category reflect a common perceptual theme; there is no requirement, how-
20
ever, that they be similar in terms of the degree of pref- perception and erence. categorization
Wonderful as such procedures are in terms of identify- ing thematic commonalities, it is important to remember that these are computational techniques. The outcome of the analysis is the designation that certain scenes "belong together," that they form a group. What it is that they have in common, however, is a matter of interpretation. The re- searcher (not the computer) provides such an interpreta- tion; others may offer alternative interpretations. For ex- ample, the researcher may examine the scenes identified by CIM as belonging to a single category and decide that their common element is the predominance of dense vegetation. Someone else examining these scenes may point out that most of them include a road. The interpreta- tion of the category would likely change with this different emphasis. However, if examination of other groupings derived from the same study reveals roads as predominant in the other groupings as well, then the interpretation based on roads is less germane. Though there is room for difference in opinion when making such interpretations, this is much less a problem than one might expect. Gener- ally, the common themes in a category are easily identified.
Two Examples
Imagine a study where individuals are shown a set of slides that depict various urban settings with which they are not specifically familiar. There are scenes of various contemporary structures that might be used as office buildings, or possibly as residence halls on a campus; there are older buildings that might be used for civic or commer- cial purposes; there are structures with tall smoke stacks that suggest industrial use, and so on. Some of the build- ings are along narrow alleys; some are surrounded by con- siderable vegetation; some are of unusual architecture, and others are nondescript. As individuals view these slides they rate each scene in terms of their preferences (5 if they like it a great deal, 1 if they don't). Quite aside from determining which scenes are more or less preferred, how would these be categorized? Would scenes with pre- sumably similar functions be grouped together, or would the age of the structure play a determining role? Is the set- ting, rather than the building, particularly salient?
21
the preference The answer we obtained in studying this question (Her- for nature zog) Kaplan, & Kaplan, 1982*) was that the groupings
could not be explained in terms of any one of these dis- tinctions. Using CIM, five categories emerged. By and large these were mixed in terms of function, except that structures that appeared to be factories did not combine with structures that were likely to have nonindustrial use. The age of the building seemed to be a salient aspect of categorization. One of the groupings consisted of jDldex buildings, varying widely in function. Another consisted of more contemporary structures. Scenes that were par- ticularly striking from an architectural standpoint also formed a separate category. The final category, however, was formed on the basis of neither age of the structure nor function. It consisted of the scenes that showed a con- siderable amount of foliage - an urban nature grouping (Fig. 1.1).
In terms of environmental perception, then, this set of results suggests that the categorization is based on several criteria: function, age, and type of architecture as well as vegetation are reflected in the groupings. For scenes that consist mostly of built elements, distinctions are apparent \ that relate to architecture and likely use. Settings that are
more natural were categorized separately, however.
Let us consider another set of pictures. This time res- idents were asked to indicate their preferences for each of 40 printed photographs of areas in and near large-scale, low-rise apartment complexes. These areas were similar to settings near where these residents lived and showed a variety of "nature" (e.g., scenes of relatively large open areas surrounded by apartment buildings, smaller natural areas between structures, areas that were neatly mowed, landscaped regions near the buildings, as well as natural areas where buildings were not present in the scene).
How is this type of environment perceived? It would seem reasonable that the underlying commonalities would be on the basis of function - what activities one might carry out. If that were the case, larger open areas where one might play ball would be grouped together by the CIM procedure. Perhaps the architecture of the buildings would be an important consideration, so that scenes of buildings that are similar in style would group together.
The results of this study (R. Kaplan, 1985a*) showed that the size of open space was not a factor in itself; nor was the tidiness or maintenance of the area. Places where one
22
Figure 1.1 Examples from four of the categories in the Herzog, Kap- lan, and Kaplan (1982*] study. Top Jeft is representative of Older Buildings category; top right, Contemporary Life; bottom left, Un- usual Architecture; bottom right, Urban Nature. For examples of the fifth category, see Figure 2.1.
could take a walk or play ball or sit and sun did not form the basis for grouping scenes. Rather, the results suggested that the basis for grouping was related to two factors: the balance between the buildings and the natural areas and the arrangement of the natural area itself. Scenes in which the apartment buildings (all of these were no more than two stories high) dominated the setting formed a distinct grouping. The fact that there were relatively large open areas and that the grass was neatly mowed did not lead these scenes to be grouped with others which also included open areas but where the buildings were more in balance with the natural setting (Fig. 1.2). At the same time, scenes in which the buildings were not visible at all formed dis-
23
Figure 1.2 All four scenes depict relatively large open areas; they were not, however, grouped in a single category [R. Kaplan, 1985a*). The scenes in the top row were included in the "faceless" category, and the bottom row scenes were part of a different category.
tinct groupings. There were, in fact, two such groupings (Fig. 1.3). One consisted of a great diversity of settings, with vegetation varying in both density and height. These can be described as "nearby-natural areas." The two scenes that formed a separate grouping both showed large mowed areas with a few trees.
Although kinds of settings represented in this study dif- fered in many respects from the previous example, the results show several interesting consistencies. (1) The categorizations in both studies are not based on any one set of criteria. (2) The issue of "facelessness" (i.e., anonymous and undifferentiated) seems to be an important dimension of categorization. In the second study, the facelessness
24
Figure 1.3 Though similar in being "natural," with no houses evi- dent, these scenes are nonetheless drawn from two distinct cate- gories (R. Kaplan, 1985a*).
was related to the similarity of the structures and their dominance over the natural elements. In the previous study, the scenes in the factory grouping communicated a "rather desolate feeling by virtue of both their scale and their facelessness" (Herzog et al., 1982*, p. 48). (3) The amount of nature in the scene is an important determinant of categorization.
By exploring the pattern of categories in these two studies, we find that each sheds light on a set of concerns that are intrinsic to the environment it subsumes. At the same time, meaningful comparisons are possible across studies despite their different kinds of settings. If we now expand this comparison to a much greater diversity of en- vironments, spanning perhaps 30 or 40 different studies, we can begin to achieve a greater understanding of how the natural environment is perceived.
25
THE PREFERENCE TYPES OF PERCEPTUAL CATEGORIES FOR NATURE
If we look at the empirically based categories that these many studies have provided, we are faced with an enor- mous analytic task. There might easily be 150 such cate- gories derived from the CIM results of these studies. Find- ing a pattern in such a collection can be a challenge. As it turns out, however, the categories can be described in terms of two major types (although some instances may fit both]. One of these types can be described as based on con- tent; the other focuses on spatial configuration.
The content-based categories have as their theme or common characteristic that they deal with specific objects or elements. Contents are relatively easy to name. Water- scapes provide a good example. In fact, "natural environ- ment" itself is such a content domain. As we have seen in the first example, in a study where the scenes are largely of the built environment, "nature" is the basis for a category. Similarly, in studies where the scenes are mostly of natural settings, the nonnatural scenes often form a distinct cate- gory (Fig. 1.4). Anderson's (1978*) study provides a further example; one of the categories in his study of forest prac- tices had in common that the scenes all depicted stands of red pine.
The spatial configuration categories are based on the way the elements are arranged in the implied space of the scene. The role of spatial organization has emerged repeat- edly in our research as a central theme in understanding people's reactions. We have gradually come to think of people's reactions as an outcome of their imagining them- selves in the setting. It is as if one asked oneself what one's possibilities would be if one were there, how easily one could move around in the scene. As individuals rate these scenes they are not, of course, likely to be aware of this process; it appears to occur automatically, effortlessly, and very quickly.
The groupings from the housing study (R. Kaplan, 1985a*) provide examples of both kinds of categories. Recall that two of the categories consisted of scenes show- ing only natural elements and no residences (Fig. 1.3). One of these, depicting scenes with smooth ground textures and a few trees, could be characterized as Parkland. This category is based on more than a content distinction. The spatial configuration plays a central role here: These set- tings permit one to see far into the scene without obstruc-
26
Figure 1.4 These scenes provide examples from four different stud- ies of categories based on human-influenced elements in the context of relatively natural settings. Top Jeft is representative of Hudspeth's (1982*) Industry category; top right, Hammitt's (1978*) Boardwalk category; bottom left, Gallagher's (1977*) Building category; bottom right, Kaplan and Herbert's (1988*) Residential category.
tion. By contrast, the scenes of the other natural-areas category consist of a diversity of nearby-natural areas without common qualities with respect to their spatial organization. Here the nature content itself seems to be the basis of the category.
It is perhaps not surprising that content is a major aspect of environmental perception. Grouping what the environ- ment contains into coherent content categories is, after all, basic to cognitive economy. By establishing such group- ings it is possible to react to a wide range of different situations in a similar way. Because one is constantly con- fronting situations that are at least slightly different from any that were previously experienced, the use of cate- gories is not merely convenient but essential if one is to apply one's past learning to new situations. At the same time, however, as will become apparent in the next sec-
27
the preference t-on> -t «g no means obvious what constitutes the for nature specific content domains in the way people experience the
environment.
The categories based on spatial configuration, however, provide a different set of insights and challenges. In these cases interpretation of the CIM results may require more than a word or two to describe the grouping. But even when the category can be defined readily (e.g., Parkland), these categories tend to be based on an implicit interpreta- tion of the opportunities and constraints afforded by the space.
It is important to realize that neither type of categoriza- tion - whether based on content or spatial configuration - necessarily reflects categories that individuals would pro- vide verbally. The perceived commonalities that lead to these categorizations are based on ratings of scenes; no verbalization is needed. There is no assumption that citi- zens in Anderson's study would necessarily call the scenes "red pines" (as opposed to pines or firs). Nor are par- ticipants likely to label scenes as "impoundments," even though such a category was derived in the storm drain study (R. Kaplan, 1977a*). Note that although the dis- tinction between these two major types of categories is helpful in most cases it must not be assumed that all categories fit clearly into one of these types.
Let us turn now to a more extensive examination of these two types. It is from such an analysis that we can come closer to understanding environmental perception and, more particularly, to seeing how people with no par- ticular expertise in this area experience the natural en- vironment.
Content-Based Categories
A major underlying theme in many of the content-based categories that have resulted from the CIM procedure con- cerns the balance between human influence and the natu- ral area. Many of the studies show a great sensitivity to kinds and degrees of such balance. In other words, the research results suggest that an underlying component of perception is a classification in terms of the degree of human influence.
As we have already seen, this was the case in the hous-
28
ing complex study (R. Kaplan, 1985a*), where three de- perception and grees or levels of balance were identified: the faceless, categorization building-dominated content; open residential areas; and nearby-natural areas with no visible residences. Berris's (1988) study, in a totally different context, showed a similar pattern, with distinct categories showing the face- less, hard spaces of buildings in a setting devoid of plant- ings as opposed to settings where the natural and built components are balanced. In Gallagher's (1977*) study, the balance between the modern commercial building and the surrounding natural area was also a strong determinant of categorization.
The results of Miller's (1984*) study were particularly in- structive in showing an underlying sensitivity to the bal- ance between human influence and the jiatural environ- ment. Participants in the study were ferry passengers departing from Vancouver, and the scenes they were asked to view depicted places along the British Columbia coast- line. Five of the seven CIM-derived groupings in this study reflected different kinds of juxtaposition between the natural and the human-influenced. These ranged from what Miller called Man-Dominated Nature (scenes of in- tensive development) to Man in Harmony with Nature (scenes with Small Structures in Natural Settings). Miller described the categories between these extremes as reflect- ing: Man-Intruded Nature (Shoreline Roads), Presence of Man in Nature (indications of logging), and Man's Past Presence in Nature (Clearcuts and Natural Bald Spots).
Quite consistently, in studies of natural settings, where most of the scenes do not include obvious examples of human-sign, those scenes that depict human habitation characteristically emerge as a CIM-based cluster (Fig. 1.4). Herbert's (1981*) study of a rural, largely natural portion of an urbanizing county in Michigan provides a good exam- ple. One of the four groupings, Residential, includes vir- tually all the scenes that show any housing at all. In CIM analyses based on Ellsworth's (1982*) study of rivers and marshes, one of the five clusters consisted of scenes that have in common the presence of human influence. These included a diversity of influences - a bridge, houses, dumped vehicles and trash - suggesting that the perceived human intervention is the likely basis of categorization here.4
It would seem then that one of the consistent bases of
29
perceptual differentiation reflects a "built" as opposed to a "natural" content, when the sampled scenes include both. How broadly the content domain is [perceptually) defined depends on a number of factors. One of these involves the frame of reference provided by the set of pictures to be judged. If relatively few scenes in a particular study reflect clear human intervention, whereas the remainder are more "natural," the few seem to cluster together even if their contents may be fairly disparate. (If such disparate con- tents were the frame of reference in a particular study, the CIM-based categories would presumably reflect the diver- sity of content rather than emerging as a single cluster.)
Another factor influencing categorization seems to re- flect the apparent importance of the domain. For example, industrial scenes are consistent in forming a category (e.g., Hudspeth, 1982*; Miller 1984*), suggesting that this is a meaningful perceptual category not only for land-use planners and realtors but also for citizens. In other cases, the absence of a category may provide indication that a content domain is not perceived as particularly salient. Roads, although easily labeled and a clear content, have only once appeared as a distinct category across many studies in which they were included. In the S. Kaplan, Kaplan, and Wendt (1972*) study, the Nature scenes in- cluded settings with unpaved roads (Fig. 1.5) and even in one instance with a parked car; in Anderson's (1978*) study, scenes with roads appeared in several groupings, suggesting that roads were not perceived as the dis- tinguishing characteristic. In the housing study (R. Kaplan, 1985a*), neither roads nor^cars (parking areas) appeared as distinct categories. (Miller's study is the exception here, as the only two scenes with a road, paralleling the shoreline and clearly cutting through an otherwise forested area, did form a category.)
Categorization is also likely to be affected by familiarity or knowledge of thejdnds of scenes included in the study. The Red Pine grouping in Anderson's study, for example, would be less likely to appear as a grouping if participants were not local citizens with a great deal of familiarity with such forests. Such familiarity may also account for the lack of grouping of "residential" areas as a common category. In the S. Kaplan et al. (1972*) study, for example, the urban and nature scenes each formed broadly defined categories. By contrast, participants were not identifying residential scenes simply as "residential." This suggests
30
PERCEPTION AND CATEGORIZATION
Figure 1.5 The presence of clear human influence did not prevent this scene and others like it from inclusion in the empirically based Nature category in the S. Kaplan, Kaplan, and Wendt (1972*) study.
that with respect to residential settings finer distinctions are characteristically made. Working-class single-family homes are not perceived as belonging to the same category as ^architecturally distinctive single-family homes, and apartments distinguish yet another cluster. (In fact, subse- quent studies of residential settings [e.g., Frey, 1981, and Widmar, 1983] corroborate this tendency toward much finer discriminations in this context.
To summarize, then, the balance between the built and the natural is a consistently dominant theme in the experi- ence of the environment. Based on empirically derived groupings, human influence emerges as a salient attribute. At the same time, however, results of these studies suggest that human intervention is not a unitary concept. In some instances, a mixture of human influences is perceived as essentially similar, whereas in other cases - mostly in the context of residential land use - human intervention is further distinguished into several distinct categories.
Spatial Configurations
Even more striking than the particular content domains, however, are the categories for which content is not the
31
the preference distinguishing characteristic. In these cases it is the spatial for nature configuration of the scenes that appears to account for the
categorization. The "space" in question here is not the two-dimensional space of the picture plane but the in- ferred three-dimensional space of the scene the photo- graph depicts.
These categories suggest that an underlying criterion in making a preference judgment is an evaluation of the scene in terms of presumed possibilities for action, as well as potential limitations. Even in the very rapidly made rat- ing, an important consideration involves an assessment of the scene in terms of what it makes possible, what it per- mits one to do. Gibson's (1979) concept of "affordance" is similar: An affordance is what an environment offers the perceiver, or, in other words, what the perceiver would be able to do in the setting. In the research discussed here the focus is on the assessment of potential actions applied to scenes. People are evaluating an entire setting in terms of potential actions - even though they cannot carry out the actions because the setting is presented photographically.
In virtually all the studies that included a variety of natural scenes, the CIM procedure led to several distinct categories that are essentially natural. The existence of separate categories within one study suggests that the presence of vegetation (as a content domain) is not the only distinguishing characteristic; other properties of the scenes must be salient in forming distinct groupings. Com- parison of these categories across studies suggests two themes as particularly prominent in distinguishing among such separate nature groupings. One of these entails the degree of openness, and the other involves spatial defini- tion.
The wide-open theme has been evident in the CIM results of many studies. The scenes comprising a Wide Open category generally lack any particular differentiat- ing characteristics, and the sky occupies a considerable portion of the scene (Fig. 1.6). Such landscapes might be of farmland (R. Kaplan, 1977a*) or of unused roadside land (Ulrich, 1974*) or of the Low Flat Shorelines (Miller, 1984*). They might also be scenes of Marshes (Ellsworth, 1982*) or the Bog mat (Hammitt, 1978*).
In Anderson's (1978*) study two separate clusters can be included under the wide-open heading. Participants in this study were local people, knowledgeable about forest prac-
32
PERCEPTION AND CATEGORIZATION
Figure 1.6 Examples of categories reflecting wide-open, undifferen- tiated spatial arrangements (from scenic roadside study, R. Kaplan, 1977a*).
tices (and dependent on forestry as the major factor in the local economy). The CIM-based results thus provide an in- dication of their perceptions of the local landscape. One of the two wide-open categories Anderson called Heavily Manipulated Landscapes. It included scenes of recent clearcuts, cutover stands, cleanup after harvest, and poor- ly stocked areas. The other category, Open, Unused Land, Anderson says "may be considered as wildlife openings, old uncultivated fields, or meadows." The fact that two separate categories emerged poses an interesting situation.
33
the preference Based on a visual analysis of the scenes, it is true that both for nature categories are generally open and lack distinctive features.
The two categories differ, however, in the land uses they reflect or, more specifically, in the degree of human in- fluence they represent. Thus, although the openness of the scenes is an indication of the role of spatial configuration, the emergence of two separate categories of this kind sug- gests that content also plays an important role in the per- ceptual process.
Scenes that lack openness also characteristically emerge as a separate category. Here too there is a lack of focus or of any differentiating characteristics, but rather than giv- ing a sense of endless open space, the view is blocked. The inability to see into the scene may be created by dense foliage, by an embankment, or by other major obstruc- tions. From an affordance point of view, either end of the openness continuum is limited. In a blocked setting, one's ability to see or move would be seriously limited. In a wide-open situation, by contrast, movement and visual ac- cess may be unlimited, but finding one's way back may be made difficult by the lack of distinguishing features.
In the second major type of spatial configuration cate- gories these problems are considerably reduced. In these instances the scenes can be characterized as having spatial definition. Space can be defined in many different ways. What they have in common involves the presence of dis- tinct edges or landmarks that help structure the setting. Ex- amination of the clusters that have appeared in the various studies we are considering here is helpful in understanding this notion.
Consistently, across many studies, scenes that are rela- tively open but have some distinct trees form a separate grouping. It is difficult to judge the depth of an undifferen- tiated surface; even a relatively small number of trees, however, contributes markedly to the spatial definition of a scene. In most studies an appropriate name of such relatively open yet spatially defined clusters might be parkJike (Fig. 1.7). As already noted, this appeared in the apartment complex study (R. Kaplan, 1985a*); similar categories were also found in the study of the storm drain (R. Kaplan, 1977a*), Ulrich's (1974*) study of roadside scenes, Hudspeth's (1982*) study of waterfront revital- ization, and Herbert's (1981*) cluster of Manicured Land- scapes in an urbanizing rural setting. In the context of less
34
Figure 1.7 Examples from categories that are relatively open, yet show clear spatial differentiation.
managed natural settings, Woodcock's (1982*) Savanna scenes also illustrate this phenomenon.
The comparison of these studies shows interesting par- allels for environments that are relatively open and for those that are generally forested. In the relatively open natural settings, two major types of categories that are based on spatial configuration appear quite consistently: the wide-open areas with little differentiation and the relatively open areas with clear spatial definition. In for- ested environments, comparable themes appear: Separate categories emerge for forests that are dense and seemingly impenetrable and for relatively open forests. In the forest, then, the dense, blocked configurations are parallel to the wide-open area. Both are characterized by the degree of openness. And both provide less spatial definition than do their better-structured counterparts.
35
Figure 1.8 Forest-based categories that differ in whether they are relatively open (top row) or more blocked (bottom row).
Several studies provide examples of these two themes in the forest context (Fig. 1.8): In Woodcock's (1982*) study, the rain forests tended to be grouped together and consti- tuted the more "blocked" landscapes, whereas the mixed hardwood forests were generally more open. In both the scenic route study (R. Kaplan, 1977a*) and a study of wilderness areas in Michigan's Upper Peninsula (R. Kap- lan, 1984b*), the degree of openness of the forest was a key basis for categorization. In Anderson's (1978*) study, the grouping that exemplified forest scenes with spatial defi- nition was the Planned Spacious Openings and Scenic Roads grouping. Here spatial definition is enhanced by the openings in the forest and, in some instances, by paths or roads. Generally, in the forest context, lack of underbrush and sufficient openness that one can see through and among the trees enhance the spatial definition.
Thus both openness and spatial definition constitute salient perceptual categories that emerge from the empiri-
36
cal results of preference judgments. They suggest that per- perception and ception entails a very rapid (albeit unconscious) assess- categorization ment of what it is possible to do in the setting. In the dense, blocked forest views, where there is considerable un- derstory or a mass of foliage, neither visual nor locomotor accessibility is apparent. The opportunities for acquiring knowledge are seriously impaired. In the relatively more open forests and the more open areas with greater spatial definition, by contrast, there is a sense that one could func- tion more effectively either because the transparency among the trees permits increased visual access or because Jhe smoother ground texture suggests that locomotion could be accomplished relatively easily. The presence of landmarks, of features that contribute spatial distinctive- ness such as a few trees in an otherwise relatively open set- ting, provides spatial definition and suggests that way- finding will be possible. On the other hand, scenes reflecting a sameness - either in a wide-open area lacking structure or in a dense woods - suggest a greater likelihood that one might get lost.
some concluding comments
There are many ways to categorize a particular environ- ment. Use of preference ratings by untrained participants (in conjunction with a Category-Identifying Methodology) yields categorizations that are distinctly different from those generated by various professionals. The groupings identified across a great diversity of studies provide some insights into the salient aspects of environmental percep- tion.
The use or function of the land is part of the implicit categorization underlying environmental perception. Thus certain kinds of uses tend to emerge as distinct clusters; in- dustrial land uses serve as a good example. But land use is also too broad a category in many cases; the empirically derived groupings indicate that neither residential land uses nor natural areas tend to be experienced as unitary content domains. Rather, these are disaggregated into separate groupings reflecting subtler differentiations. In fact, the balance between the built and the natural is an im- portant basis for such differentiation.
Content is only part of the basis for the way the environ- ment is perceived. People are extremely sensitive to the
37
the preference spatial properties of the environment. It is apparent that in for nature the rapidly made and largely unconscious decision regard-
ing preference, there is an assessment of the glimpsed space and its qualities. This rapid assessment appears to be heavily influenced by the potential for functioning in the setting. Thus indications of the possibility of entering the setting, of acquiring information, and of maintaining one's orientation emerge as consistently vital attributes. The consistent presence or absence of these attributes fre- quently provides a basis for the formation of a coherent perceptual category. Wide-open, undifferentiated vistas and dense, impenetrable forests both fail to provide infor- mation about one's whereabouts, and both consistently ap- pear as distinct perceptual categories. On the other hand, scenes that convey a sense of orderliness (such as parklike and manicured settings) also tend to form a distinct cate- gory. In such settings, the smooth ground texture affords prediction about how one could function in the setting. Similarly, forest's that are more transparent, with light filtered through the trees and with suggestions of paths, provide information about accessibility and direction. Thus a major underlying basis implicit in these categori- zations is an assessment of the possibility of functioning in the setting.
It is striking how profoundly information-based these categorizations have been. The majority of the expert- generated category systems have little to do with the way people process information. Results of the various studies summarized here suggest that what people experience as salient in the landscape involves informational patterns, readily interpretable in terms of requirements for adaptive behavior. The way space is organized provides informa- tion about what one might be able to do in that space. A relatively brief glance at a scene communicates whether there is room to roam, whether one's path is clear or blocked.
Thus there appear to be both an empirical and a theoreti- cal basis for categorizing landscape scenes. As is often the case with a satisfying research experience, these cate- gories would have been hard to anticipate but in retrospect make intuitive as well as theoretical sense.
38
1. Levin's (1977*) study was an early effort that showed strong similarities in reactions to on-site and photographic represen- tations of the same environments. Shuttleworth (1980) and Ulrich (1979) discuss some of the distinct advantages of using photographs. Zube, Simcox, and Law (1987) provide a review of some of this literature and also argue for its validity and reliability. Kellomaki and Savolainen's (1984) study, carried out in Finland, is noteworthy because they used both "or- dinary city dwellers" and students trained in forestry to study the differences between field and laboratory ratings of forest landscapes. Coeterier (1983) reports that photographs can be substituted for actual settings more reliably for "small-scale landscapes" than for "large-scale landscapes with micro- relief."
2. Categories and related concepts (e.g., prototypes, schemata, natural categories) play a central role in modern cognitive psychology. The findings are generally consistent with the approach we have taken, although much of the work in the area (e.g. Rosch, 1978; Tversky & Hemenway, 1983) has been (from our perspective) hampered by the traditional informa- tion-processing emphasis on alphanumeric material as op- posed to drawings and photographs. The most recent trend in this area is away from the strict logical definition of a category and toward concepts with fuzzier boundaries. Some researchers feel that even the idea of fuzzy boundaries is in- sufficient to account for the subtlety of category use and argue instead for the centrality of some sort of implicit theory a person calls on to decide whether something does or does not belong in a category. For stimulating discussions of re- cent thinking about categories, see Lakoff (1987); Murphy and Medin (1985); and Posner (1986).
3. CIMs previously were designated content-identifying meth- odologies in S. Kaplan (1979a). The abbreviation has not changed, but category is adopted here as a more descriptive name for the procedure. Since some categories are based on spatial configuration rather than on content, this confusion is avoided.
4. Though these scenes received relatively low preference ratings, it was not the distaste for the intervention that was the basis for the perceptual grouping. The marsh scenes were no better liked but were a clearly distinct theme. In other words, the categorization is based on a similarity in reaction to a group of scenes, not necessarily on whether they are liked or disliked. Signs of blatant human influence in an otherwise natural setting are perceived as a distinct category.
39
THE PREDICTION OF PREFERENCE
There is no question that people's tastes differ. With respect to culinary tastes, the diversity is evident around the dinner table at home or in a restaurant. With respect to clothes, differences in preference are also readily apparent. Even with respect to natural environ- ments, the variation in preference is quickly evident. Con- sider pictures and posters of natural places that people dis- play on their walls. There are the grand, the intimate, the awe-inspiring, the tranquil, the ephemeral, the common- place, the wild, the tame, the representational, and the abstract. A look around a custom frame shop provides a sampling of this diversity.
It is easy, however, to draw false conclusions from such diversity. One frequent generalization is that "there is no accounting for people's tastes" or that "beauty is in the eye of the beholder." These statements suggest that taste is random, as variable as people are. Another false inference is that tastes are frivolous or whimsical or that the issue of preference is simply not of great moment. It might be nice to have a vase full of fresh flowers, but one could certainly survive without.
If indeed tastes were random and the question of prefer- ence were of no great consequence, there would be little reason for this chapter. The study of environmental prefer- ence, however, shows remarkable consistency, despite demographic differences and across diverse settings. Giv- en these results, there is reason to suspect that environ- mental preferences provide a glimpse into some essential ingredients of human functioning. The purpose of the present chapter, then, is to explore the question of prefer- ence for natural environments in a relatively generic sense - in terms of attributes that are likely to be pertinent across
40
the diversity of humankind. (In chapter 3 we will return to prediction of the question of differences in preference.) preference
PERCEPTION AND PREFERENCE
The discussion of perceptual categorization in chapter 1 was based on ratings of scenes in terms of how much the viewer liked them - in other words, preferences. As we de- scribed, Category-Identifying Methodology (CIM) pro- vides an approach to analyzing these ratings by extracting common patterns in the way people respond. By compar- ing the categories that emerged in numerous studies we found that these fall into two major types: content-based and spatial configurations. In chapter 1 these types of categories were examined as a way to understand how the environment is experienced. Preference ratings were used as a vehicle to understand perception - but the preferences per se were not included in the discussion.
Perception is quite obviously important to survival. It is, for example, clearly adaptive to be able to perceive danger. But being able to perceive what is safe and what is dangerous is not enough. If the information an organism acquires through the power of perception is to aid in its survival, it is essential that it not only perceive what is safe but also prefer it (Appleton, 1975; S. Kaplan & Kaplan, 1982). Such a pattern of preferring a suitable habitat over an unsuitable one is widespread in the animal kingdom and may well be characteristic of humans as well.1 If this were the case, one would expect that what is basic to per- ception must also be important to preference.
It is time, then, to look at the perceptual categories to see what they reveal about preference. To do this we will use the same studies that were the basis of the discussion in the preceding chapter (identified by * in the discussion here and summarized in Appendix B). Recall that participants in a particular study are asked to view a series of scenes (photographs or slides) and to indicate their preference for each one, using a 5-point rating scale. These preference ratings are analyzed using CIM procedures (see Appendix A), resulting in a number of categories that reflect com- monalities in rating patterns. The categories tell us how the environment that is presented in the scenes is per- ceived by the participants. To determine how much each
41
the preference 0f these perceptual categories is liked, the preference for nature ratings for the scenes comprising each category are av-
eraged. These mean preference values are the basis of the analysis here.
Before examining these preferences in terms of the dif- ferent types of categories, it is useful to provide some sense for the mean preference values that are characteris- tic in such studies. The groupings typically consist of any- where from 3 to 12 scenes. For an average preference rat- ing of a grouping to be near the extremes of the 5-point scale would require great unanimity. This is unlikely not only because people vary but also because the various scenes that comprise a perceptual category will differ in details that can affect the degree of preference. In fact, it is most unusual to find a grouping with a mean of 4.0 or higher or one that is at 2.0 or lower. In general, means below 3.0 reflect "low" preference, and those at 3.7 and above can be considered relatively "high."
In addition to considering these absolute levels, it is im- portant to consider the ratings relative to other category means within a given study. Given the very diverse con- texts for the studies, the range of ratings differ somewhat. For example, in the study of the nearby storm drain system (R. Kaplan, 1977a*), ratings tended to be relatively low in general - perhaps because most of the scenes depicted near-home places that needed some care. The most pre- ferred scene in the study was from another area entirely, unknown to the participants and extremely well main- tained. By contrast, in Hammitt's (1978*) study, ratings tended to be relatively high. In this situation, people were visitors to a featured attraction ( a bog trail at Cranberry Glades in West Virginia), and most of the scenes were taken in that setting. Here the scenes taken elsewhere generally received lower ratings.
Content-Based Categories
As discussed in chapter 1, the content-based categories that were empirically derived across the various studies reflected a concern for the balance between human in- fluence and natural areas. In studies that included indus- trial scenes these invariably emerged as a category. Not surprisingly, these categories were also consistently low in preference. In both Hudspeth's (1982*) and Miller's (1984*)
42
studies, for example, the industrial groupings received by prediction of far the lowest preferences (mean of 1.8 and 2.1, respec- preference tively). The Alley/Factory grouping in the Herzog et al. (1982*) study, with a mean of 1.6, included a mixture of in- dustrial and other barren-seeming settings (Fig. 2.1).
It is not only the industrial scenes that receive low pref- erence ratings. In the vast majority of studies, the cate- gories that reflected the clearest human influence were relatively lowest in preference. The human influences varied considerably in type. In the S. Kaplan et al. (1972*) study, it was the Urban that formed a single category (mean 2.2). In the multiple-family housing study (R. Kap- lan, 1985a*), the Building-Dominated group (mean 2.2) (shown in Fig. 1.2, top row) reflected the strongest sense of human influence. Herbert's (1981*) study included a rural Residential category; in this case the buildings were far less dominant in the landscape, but relative to the other
Figure 2.1 Two scenes from the Alley/Factory category in Herzog, Kaplan, and Kaplan (1982*) study.
43
the preference scenes in the study they showed the strongest human in- for nature fluence, reflected in a mean rating far lower than any
others in the study (2.8). The Human Influence category based on Ellsworth's (1982*] data showed a great mixture of intrusions, but the mean rating was similarly low (2.5).
Anderson's (1978*) study is striking for having a strongly unpreferred category (mean 2.3) where the human in- fluence involves no built elements. Rather, these scenes show the results of clearcutting; Anderson refers to the category as Heavily Manipulated Landscapes. Given that the viewers in this case were local residents who are highly familiar with such scenes and whose economy depended on the lumber industry, the finding is even more note- worthy.
Not all the human-influence content categories, how- ever, were low in preference. Hudspeth's (1982*) Boats category received midrange ratings (3.6), and Miller's (1984*) Small Structures in Natural Settings category was second highest among the categories in his study at 3.9. Hammitt's (1978*) Boardwalk category (included in Fig. 1.4) was high in preference (mean 4.1). Here a wooden boardwalk was the common theme in four scenes that were otherwise totally natural. In both the Miller and Hammitt instances, the human influence is central to the content, but the built component is in keeping with the set- ting and does not dominate the natural elements in the scenes. Figure 2.2 provides a summary of the relative pref- erences of these categories, based on 10 studies.
At the other end of the continuum, the most preferred content-based categories have generally been ones where nature is dominant in the scenes. The two studies used as examples in chapter 1 provide good examples: In the multiple-family study (R. Kaplan, 1985a*) the Nature category received a mean of 4.0. In the Herzog et al. (1982*) case, the Urban Nature group was by far the most pre- ferred (3.6, two scale points higher than the Alley/Factory grouping). In many other studies the means have ranged between 3.5 and 3.7 - always significantly higher than for the categories that were relatively less natural within the same study. In analyses based on Ellsworth's (1982*) data, for example, though all scenes showed waterscapes, the category with snow-peaked mountains in the background was a full scale point higher in preference than the one with human intrusions.
44
42 PREDICTION OF
4.1 Boardwaik (Hamm*. ,978-, PREFERENCE
3.9 Small structures in nature setting (Miller, 1984*)
3.8
3.7
3.6 Boats (Hudspeth, 1982*)
3.5
3.4
3.3
3.2
3.1
3.0
2.9 Building (Gallagher, 1977*)
2.8 Rural residential (Herbert, 1981*)
2.7
2.6
2.5 Human influence (Ellsworth, 1982*) 2.4
2.3 Heavily manipulated (Anderson, 1978*)
2.2 Urban (S. Kaplan et al., 1972*); Building-dominated (R. Kaplan, 1985a*) 2.1 Industrial (Miller, 1984*)
2.0 1.9
1.8 Industrial (Hudspeth, 1982*) 1.7
1.6 Alley/ Factory (Herzog et al., 1982*) 1.5
Figure 2.2 Relative preferences for categories reflecting human in- fluence in context of scenes that are mostly natural.
Spatial Configurations
In chapter 1 we described the second type of perceptual categories as reflecting the interpretation the observer in- tuitively and unconsciously makes about the ability to function in the pictured "space." These interpretations are based on how the space is organized. Recall from that dis- cussion that two components are particularly salient in these perceptual categories: degree of openness and spa- tial definition. We suggested that the preference judgments seem to reflect a very rapid assessment of what one could do were one in the scene. The perceived openness (either wide-open or seemingly blocked) seems to be critical to these rapid assessments. Similarly, a sense of depth or focus affects the judgment about how readily one could function in the pictured setting.
One of the most consistent categories to emerge across the various studies involves areas that are very open, with the sky taking up a large portion of the scene and a lack of distinctive features in the foreground (Fig. 1.6). These
45
Figure 2.3 Examples from the two categories in Anderson's (1978*) study that reflect wide-open configurations. Scenes in the top row are taken from the Open, Unused Land category, and the bottom row scenes are from the Heavily Manipulated Landscapes category.
groupings are generally among the lowest in preference within any particular study, with preferences ranging from 2.3, in the case of the Marsh grouping based on Ellsworth's (1982*) data, to 3.1 in Herbert's (1981*) Pastoral grouping. Scenes of distant farmland, both in the scenic highway (R. Kaplan, 1977a*) and in the R. Kaplan, Kaplan, and Brown (in press*) studies, were similarly low in preference, with means of 2.7 and 2.9, respectively. The Bogmat scenes in Hammitt's (1978*) study are exceptional for their con- siderably higher preferences (mean 3.7), although this category was distinctly less preferred than the other cate- gories to emerge in that study.
Recall from the previous chapter that in Anderson's (1978*) study two categories emerged that fall under this "wide-open" heading. Here spatial configuration and con- tent considerations both play a role, as the two categories
46
are distinguishable in terms of the degree of human intru- sion (Fig. 2.3). As already mentioned, timber harvesting is an essential aspect of the local economy for participants in Anderson's study, and yet these residents far preferred the scenes of Open, Unused Land to those depicting the out- comes of relatively recent harvesting (means 3.2 and 2.2, respectively).
Though extreme openness is not a preferred characteris- tic, the other extreme fares no better. Such categories con- sist of blocked views or very dense vegetation (included in Fig. 1.8), suggesting that both visual and locomotor access would be difficult. Woodcock's (1982*) study included two such categories: Dense Hardwood Forest and Rain Forest (means 3.0 and 2.8, respectively). Gallagher's (1977*) prai- rie restoration study also yielded a "blocked" grouping, consisting of the tall grasses (mean 2.5) (Fig. 2.4).
PREDICTION OF PREFERENCE
Figure 2.4 Examples of blocked views taken from Gallagher's (1977*) study. These are scenes of the Prairie planting.
47
the preference jn terms of preferences, the spatial configuration that for nature consistently generates favorable responses involves areas
that are open, yet defined. Characteristically, these have relatively smooth ground texture and trees that help define the depth of the scene (Fig. 1.7). Such categories can be called parkJike or woodlawn or savanna. Mean prefer- ences for these are always among the highest, ranging be- tween 3.7 and 4.2 (e.g., Wookcock's [1982*] Savanna, An- derson's [1978*] Planned Spacious Openings, Gallagher's [1977*] Complex, R. Kaplan's [1985a*] Parkland, Herbert's [1981*] Manicured Landscapes, and the Woodlawn group- ing in R. Kaplan et al. [in press*]. Ulrich's (1974*) study also included a clear example of such a category, but because he used a 6-point scale the preference ratings are noncomparable; nonetheless, the preferences for the park- like scenes were by far the highest. Figure 2.5 provides a summary view of these relative preferences, based on 10 studies. (See also Ulrich, 1986, and Schroeder and Green, 1985, for further studies showing the strength of prefer- ence for such configurations.)
4.4
4.3 Open forest (R. Kaplan, 1984*) 4.2 Parkland (R. Kaplan, 1985a*) 4.1
4.0 Complex (Gallagher, 1977*) 3.9 Manicured (Herbert, 1981*)
3.8 Planned spaciousness (Anderson, 1978*); Parklike savanna (Woodcock, 1982*)
3.7 Woodlawn (R. Kaplan et al.,in press); Bogmat (Hamitt, 1978*) 3.6
3.5 3.4 3.3
3.2 Open, unused (Anderson, 1978*)
3.1 Pastoral (Herbert, 1981*)
3.0 Dense forest (Woodcock, 1982*) 2.9
2.8 Rain forest (Woodcock, 1982*)
2.7 Distant farmland (R. Kaplan, 1977a*) 2.6
2.5 Prairie (Gallagher, 1977*) 2.4
2.3 Marsh (Ellsworth, 1982')
2.2 Heavily manipulated (Anderson, 1978*) 2.1
2.0
Figure 2.5 Relative preferences for categories reflecting spatial con- figuration.
48
Summary
PREDICTION OF PREFERENCE
The perceptual categories are important for identifying characteristics that are salient in the way the environment is experienced. At the same time, they also provide insight into patterns that are liked and disliked. In general, the content-based categories suggest that in the context of natural environments those with human intrusions are less preferred and those where nature dominates the built ele- ments receive much more favorable response. In other words, people are sensitive to the way human influence is positioned in the context of the natural setting. All too often these human influences are intrusive. But human in- fluence can be incorporated in a way that is responsive to the form and characteristics of the natural features.
The categories based on spatial configurations suggest that the perceptual process is sensitive to the relative open- ness and spatial definition of the setting. Preferences for the resulting categories show that preferred settings are those where it is easiest to extract the information needed to function. In very open and in blocked areas it is more difficult to anticipate what might happen, and preferences tend to be relatively low. In spatially defined areas and in open forests, by contrast, it is far easier to judge where one can venture safely and what to expect. Such categories tend to be highly favored.
THE PREFERENCE MATRIX
The empirically derived perceptual categories provide one way to examine environmental preferences. Let us turn now to a different way of looking at the preference ratings. This approach is by far the most straightforward, low-tech way to explore research results. Rather than looking at commonalities in patterns of ratings (and relying on CIM "technology"), this procedure simply examines the mean preference rating for each scene.
Using the same set of studies (all of which included 5- point preference ratings), an average rating is computed for each scene in a particular study. One then selects the most and least preferred scenes for closer examination. (The decision as to what constitutes "most" and "least" can be based on some arbitrary cutoff such as "means greater than 4.0 and less than 2.0" or the top and bottom six
49
the preference to eight scenes.) When spreading these scenes out before for nature onej jn many cases the interpretation of preference is
readily apparent.
Using this procedure, as with the CIM-based analyses, the content of the scenes often appears as the underlying basis for the different preferences. Waterscapes are often among the highly preferred settings (except, of course, in studies that consist mostly of waterscapes, such as the storm drain study [R. Kaplan, 1977a*] and Ellsworth's [1982*] rivers and marshes study). Since industrial scenes are so consistently down-rated, they are often in the least preferred pile.
Here again "content" is not a sufficient basis for un- derstanding the preferences. When looking at the most and least preferred scenes one often finds some other charac- teristics that they have in common, although these may be more difficult to express. It is such "eyeballing" that in- forms intuitions. By looking at highly preferred and un- preferred scenes we came up with some hunches about preferences. These hunches inspired the next studies, which led to revised intuitions. Gradually a framework developed that we have called the Preference Matrix. With further research and with efforts to use the framework, it has continued to evolve.2 We cannot promise that the pro- cess is yet complete. Let us turn, then, to its current rendition.
Preference and Human Needs
As we said earlier, human functioning depends on infor- mation. People seem to be extremely facile in their ability to extract information from the environment. Even the very briefest glimpse of the passing landscape provides in- formation. This information does not depend on posted signs or neon lights. It is far subtler and generally not a part of one's awareness.
The Preference Matrix is divided into two domains rep- resenting two critical facets of people's relationship to in- formation. We will first consider these two major domains, each of which has two parts, and then look at the features of the environment that enhance each of the four cells in the matrix.
The first domain involves two major categories of hu- man needs: understanding and exploration. The need to
50
understand, to make sense of what is going on, is far- reaching in its expression. Even reasonable, kindly people can become hostile and angry when they cannot com- prehend material that seems to be necessary to function- ing. In fact, there are times when such frustration is readily apparent even when the failure to understand would seem to have little immediate consequence. Occasionally one can witness such reaction when individuals view "mod- ern" art that fails to make sense to them. The hostility evi- dent at many public meetings may also be a reflection of the citizens' inability to understand the technical material thrust upon them.3
Understanding the environment, like anything else, is at least partially dependent on prior experience. Many ani- mal signs, such as nests or lodges or hives, may make little sense to one individual and yet be readily recognized or in- terpreted by someone with experience. Similarly, the en- vironmental patterns associated with different forest prac- tices may be interpreted very differently depending upon one's ability to understand them. Buhyoff, Leuschner, and Wellman (1979) have shown that preferences for trees that are diseased (orange-brown foliage from southern pine beetle damage) are higher when people are not knowledge- able about the cause of the coloration. Similarly, Ameri- cans looking at Australian landscapes are not likely to un- derstand that trees without leaves are diseased rather than showing seasonal variation, because they are not aware that Australia has no native deciduous trees (R. Kaplan & Herbert, 1987*).
Despite differences in background knowledge, however, the need to understand is quite general, jmd certain en- vironmental attributes play a strong role in making such comprehension more likely. If humans have a strong need to understand, it seems reasonable that preferences would be greater when comprehension is facilitated. We will look at some of these attributes shortly.
Though understanding is important, it is often not suffi- cient. People also prefer circumstances that require them to expand their horizons, or at least circumstances where such enrichment is a possibility. The second category of human needs, then, is the need to explore, to find out more about what is going on in one's surroundings. Exploration is an important element in accumulating experience. It in- clines one to expand one's knowledge as well as to in- crease one's capacity to understand previously confusing
51
the preference situations. At the same time, it provides a way to deepen for nature one's grasp, by inquiring into new facets of a familiar
situation.
Exploration, like understanding, is also greatly affected by previous experiences. As is true with understanding, however, even in the absence of prior knowledge, the likelihood that one will venture forth and seek more infor- mation can be strongly affected by environmental attri- butes. If the need to explore is a pervasive human need, it is also reasonable that preferences would be greater where exploration is facilitated.
The second domain of the Preference Matrix involves the degree of inference that is required in extracting the needed information. It may be easiest to think of this factor in terms of the two-dimensional and three-dimensional as- pects of the visual environment.
The two-dimensional aspects, or the picture plane be- fore one, involve information that is immediately avail- able. It is perhaps easiest to think of this in terms of a photograph of any given landscape. Consider the patches of light and dark areas distributed across the paper on which the photograph is printed. This pattern or organiza- tion of light and dark on the photograph constitutes the basis of this level of analysis. Processing information at this level requires very little inference.
The three-dimensional pattern of the actual or depicted space requires greater inference. When we consider the depth of the scene, the space unfolds before us. A setting with partially obscured elements, or with features that mark the distance available to the viewer, provides infor- mation that requires more interpretation. As we saw in chapter 1, the three-dimensional pattern invites one to imagine oneself in the scene. Thus there is greater em- phasis on what might be seen from a different vantage point as opposed to what is immediately apparent.
Informational Factors in Preference
The Preference Matrix, up to this point, has been con- cerned with the two basic informational needs - un- derstanding and exploration - and with a dimension that considers how readily available the information is. The combination of these two domains yields four distinct combinations, or patterns (Table 2.1).
52
Table 2.1. The Preference Matrix prediction of
PREFERENCE
|
Understanding |
Exploration |
|
|
Immediate |
Coherence |
Complexity |
|
Inferred, predicted |
Legibility |
Mystery |
These combinations, or informational factors, are given names here that have been part of the landscape assess- ment literature. It is important to point out, however, that these factors have been defined in many ways. Our usage is strongly affected by the context of the matrix. In other words, we see these factors in terms of the way they help one understand or explore, and in terms of a two- or three- dimensional analysis of the scene. Let us turn, then, to an examination of the environmental attributes or the charac- teristics of the way the environmental scene is organized that help define each of these four combinations.
Complexity. Two very different disciplinary traditions have focused on this informational factor. Psychologists interested in the area of aesthetics for many years concen- trated on the role of complexity. Using artificially con- structed stimulus patterns, the basic finding of this "exper- imental aesthetics" work was that people prefer patterns that are at neither the high nor the low end of the complex- ity continuum (Day, 1967; Vitz, 1966).
The other tradition that has focused on complexity dif- fers in two important respects. The emphasis has been on the environment, as opposed to artificial stimuli, and the work has not been based on research. Many of the visual resource management systems that federal land agencies have adopted (e.g., U.S. Bureau of Land Management, 1980; U.S. Forest Service, 1974) have assumed that diver- sity is an important component of scenic quality; settings with greater diversity receive higher ratings in these scor- ing systems.
In the studies under consideration here, Complexity is defined in terms of the number of different visual elements in a scene; how intricate the scene is; its richness. It thus reflects how much is going on in a particular scene, how much there is to look at - issues that call upon the picture plane, as opposed to depth cues. Clearly, exploration is en-
53
the preference Table 2 2. Relationship between Coherence and for nature Complexity
Complexity
|
Low |
High |
|
|
Coherence |
||
|
Low |
Not much there |
Visually messy |
|
High |
Clear and simple (boring) |
Rich and organized |
hanced when there is more variety in the scene, when there is the suggestion that there are more different things avail- able. It could be argued that Complexity provides content or things to think about.
Coherence. This informational factor has received rela- tively little study. Coherence, in our perspective, helps in providing a sense of order and in directing attention. A coherent scene is orderly; it hangs together. Coherence is enhanced by anything that helps organize the patterns of brightness, size, and texture in the scene into a few major units. Such features as repeated elements and uniformity of texture, both examples of redundancy, help to deline- ate a region or area of the picture plane. Properties of a scene such as texture as well as the size and location of the various uniform areas are assumed to be the province of the "location system" (S. Kaplan, 1970; S. Kaplan & Kap- lan, 1982), an ancient neural structure that processes visual information with great speed and little need for inference.
As is true with Complexity, Coherence involves rela- tively little inference, relying on the two-dimensional as- pect of the setting. A coherent setting contributes to one's ability to make sense of the environment. An orderly situa- tion is easier to understand; it may not be more inviting of exploration. The trade-offs between Coherence and Com- plexity are interesting to consider. One is tempted to con- sider a messy setting as overly complex. More likely, given this framework, it lacks in Coherence. It is important to realize, however, that a scene can be high in Complexity and in Coherence at the same time (Table 2.2).
54
Legibility. Kevin Lynch introduced this concept in his pro- found little book, The image of the city:
This book will consider the visual quality of the American city by studying the mental image of that city which is held by its citizens. It will concentrate especially on one particular visual quality: the apparent clarity or "legibility" of the cityscape. By this we mean the ease with which its parts can be recognized and can be organized into a coherent pattern. (Lynch, 1960, pp. 2 - 3)
Lynch's notion of legibility includes all of what we are calling understanding, since he points to the necessity of both coherence and structure. In his discussion of legibil- ity, however, he showed the centrality of orientation and way-finding to the ability to build a mental map of the set- ting. We have adopted Lynch's term to refer to this more structural (and inferential) aspect of understanding.
As we define it, a legible space is one that is easy to un- derstand and to remember. It is a well-structured space with distinctive elements, so that it is easy both to find one's way within the scene and to find one's way back to the starting point. It is also important that the objects be identifiable (e.g., Herzog, 1984*, 1985*, 1987*) and the scene be experienced as interpretable. (The concept of "fit- tingness," as used by Wohlwill and Harris, 1980, may similarly reflect the roles of identifiability and inter- pretability.) Legibility thus entails a promise, or prediction, of the capacity both to comprehend and to function effect- ively.
Landmarks and regions (concepts for which Lynch pro- vided perceptive analysis) that are distinctive and mean- ingful are important aids in achieving legibility. These help in building a cognitive map, in the "memorability" of a scene (or city, for Lynch). Such distinctive elements or regions are particularly useful in giving one the sense that one could both comprehend and move effectively within the setting.
Mystery. This informational factor also involves promise, but here it is the promise that one could learn more. Some- thing in the setting draws one in, encourages one to enter and to venture forth, thus providing an opportunity to learn something that is not immediately apparent from the original vantage point. There are several ways that scenes or settings can suggest that there is more information
PREDICTION OF PREFERENCE
55
the preference available. Some classic examples include the bend in the for nature path and a brightly lit area that is partially obscured by
foreground vegetation. Partial obstruction, often from foliage, and even modest land-form changes can enhance this sense of Mystery.
Thus, for Mystery (as defined here) to be present, there must be a promise of further information if one could walk deeper into the scene. This necessarily implies that it would be possible to enter the scene, that there would be somewhere to go. It is important to contrast Mystery with surprise. A path leading to a visible closed door suggests surprise but not Mystery. For the latter, the change in van- tage point needs to provide information that is continuous with what is already available, rather than a surprise. Given the continuity, one can usually think of several alternative hypotheses as to what one might discover - in other words, there is both inference and a sense of explora- tion.
Designers have long recognized the powerful effect of these qualities. Hubbard and Kimball used the word mys- tery in their 1917 text, An introduction to the study of landscape design. Their use of it extended beyond the re- stricted sense intended here. For example, their example of mystery, when "the sheer multiplicity of detail prevents our clear comprehension of the landscape, as when we look at the misty leaves and branches of a thick deciduous Wood in early spring," would probably be a better example of Complexity as used here. However, they provide many useful examples of situations that suggest Mystery. For example, they write of the "inability to see the landscape with any distinctness, as for instance when the scene is shrouded in haze or in a snow storm or in darkness." Another of their examples - "the foreground is clearly seen, but ... an important part of the landscape known to be present is nevertheless concealed, as where a river or a road winds out of sight behind some intervening barrier" - corresponds closely to Mystery as we define it. Thus we would agree with them that "the effect of mystery is the result of impossibility of complete perception," but not all instances of incomplete perception qualify for Mystery. The strong role Mystery has played in preference supports their assertion that "it is a pleasant challenge to the imagination which sets the observer to trying to determine for himself by closer investigation what is concealed from his first glance, or if this be impossible, to filling in and
56
completing the unseen landscape according to the play of prediction of his own fancy" (Hubbard & Kimball, 1917, p. 82). preference
Simonds (1920) also wrote of mystery, which he equated with curiosity:
The interest in any view is increased by an arrangement that piques one's curiosity. In illustration of this, think of woods into which one gets glimpses leading to unknown depths, bays of lakes disappearing behind islands or promontories, lawns partly hidden by projecting groups of shrubs. These give possible op- portunities for making discoveries, and such opportunities com- pete with variety in giving spice to life. The shape of a tree, the graceful or strong arrangement of its branches, the outlines and texture of its leaves, the color and forms of flowers, the curves of the earth's surface, the reflections in water - are all objects of in- terest and beauty, but beyond all these in making a view interest- ing are the elements of curiosity and mystery. (Pp. 13-15)
Cullen's (1961) use of mystery - "where anything could happen or exist, the noble or the sordid, genius or lunacy" (p. 51) - does not match ours. However, his discussion of "those aspects of here and there in which the here is known but the beyond is unknown" does partially overlap with our concept. In particular, his use of anticipation recognizes both the inference and the curiosity that are in- volved in the promise of further information.
Although designers have written of Mystery, little em- pirical work has been done to ascertain circumstances in the natural setting that might enhance it. Gimblett, Itami, and Fitzgibbon (1985*) studied this question and reported that several factors were particularly effective in enhanc- ing Mystery: screening, enclosure, physical accessibility (for example, a defined path), forest illumination, and a relative lack of "distance of view."
Perspective on Preference Matrix
The Preference Matrix evolved from extensive examina- tion of the scenes that are most and least liked across numerous studies. It was through such observations that we were led to see that Complexity, the object of consider- able study in nonenvironmental contexts, was not power- ful in explaining preference for the natural environment. The most preferred scenes in many studies, however, reflect what we have called Mystery: There is partially hidden information, and something in the scene tempts
57
the preference one to explore further. Smooth ground textures with for nature widely distributed trees that mark depth tend to be high in
Legibility and are also among the preferred scenes. By con- trast, disliked scenes often show relatively JittJe Coherence or Complexity.
The pattern that such observations suggest is that prefer- ence calls for at least a modicum of the qualities that per- mit immediate processing. A Jack of Coherence makes it difficult to understand what is before one; a Jack of Com- plexity diminishes one's likelihood of becoming engaged in viewing. It is not necessarily the case, however, that preference is enhanced by having increasing amounts of these informational factors. For the two factors that rely on greater inference, however, there is the suggestion that "the more the merrier." With more Legibility, confidence is enhanced that the setting will continue to be understand- able. More Mystery entices one to further exploration.
There is no attempt here to say that the Preference Ma- trix accounts for all aspects of environmental preference. Rather, it provides a framework that seems to be widely applicable. It points to the need for considering several factors as well as their combinations. What may seem to be a more parsimonious model, such as a continuum ranging from unity to diversity with the suggestion that most pre- ferred areas fall somewhere along it, is far too simplistic. At the same time, a lengthy list of issues that pertain to scenic quality without any organizational guidelines is also not likely to be fruitful. Consideration of the needs to understand and to explore, however, provides substantial structure and permits analysis of preferences despite con- siderable human variation.
THE EFFECTIVENESS OF THE PREDICTORS
There are many ways to approach the "prediction" of pref- erence - the question of whether one can reliably an- ticipate attributes that enhance people's preferences for particular settings. The two forms of analysis we have dis- cussed so far - one based on preferences for perceptual categories and the other involving examination of the scenes with the highest and lowest preferences - have been helpful in suggesting some issues that underly prefer- ence. These analyses have led to insights about content concerns, openness and spatial definition, and the factors
58
in the Preference Matrix as potential indicators of scenic prediction of quality or environmental preference. preference
The potential indicators, however, are derived from the interpretation of the results. In the discussion so far they have not been tested in their own right. We turn now to a third way to use the preference ratings in attempting to un- derstand environmental preference. For this approach, the potential indicators are directly examined to see whether they do, in fact, help account for the preference ratings. To do this, each of the specific attributes considered to be an appropriate candidate for accounting for preference needs to be evaluated or rated for each scene in the study. These ratings are then related to the preference ratings.
In hindsight, one's research program would be quite dif- ferent than it was. From one's current perspective, it would have been nice to have settled on a set of potential predictors or indicators, agreed on their definition, stan- dardized the procedure for obtaining ratings, and then sys- tematically studied a variety of environmental contexts. To the extent that following such an approach is an ex- pression of methodoJogicaJ preference, it would seem to emphasize Coherence at the expense of the other aspects of the Preference Matrix! The actual direction this re- search program has taken can hardly be described as a straight and narrow path; it is an expression of the power of exploration in the search for understanding. It continues to be a framework that guides exploration, with no assur- ance of where the trail will end.
EarJy Research
Our first study in the environmental preference area (S. Kaplan et al., 1972*) was based on 56 scenes, carefully se- lected to reflect a continuum ranging from natural (with minimal evident human intrusion) to strongly human- influenced (with no evident natural elements). In addition to this major focus on content, the study was oriented to examining the role of the one factor that had been em- phasized in the experimental aesthetics area, namely Com- plexity. Study participants were asked to rate each of the 56 scenes both in terms of this informational factor and to indicate how much they liked it. The correlation between these two ratings was .37, based on all the scenes.
Using the CIM approach, two perceptual categories
59
the preference emerged that were strongly content-based. The scenes for nature with clear human influence and virtually no natural ele-
ments formed the Urban category; those with natural ele- ments (including many that showed human influence) formed the Nature category. (As mentioned in chapter 1, the scenes that showed residential arrangements did not form a clear content category.) Preferences for the two content domains were strikingly different, with the natural scenes far preferred. Clearly, content was an important element in preference.
Examination of the most and least preferred scenes pro- vided further indication of the importance of content, as the most preferred scenes were of natural settings and the least preferred were urban. Examination of the most and least preferred scenes also made it evident that other fac- tors should be taken into account besides Complexity. The very first version of the Preference Matrix was coming into being. Mystery and Coherence seemed to be important factors in explaining the patterns of preference.
We therefore used the same slides again and asked separate groups of participants to rate the scenes in terms of the two new informational factors that had seemed salient: Mystery and Coherence. (Legibility was not yet part of the Preference Matrix.) For purposes of replication, and because we were concerned about the potential bias- ing effect of rating scenes with respect to both an infor- mational factor and preference, we also asked two other groups of participants to rate the scenes in terms of Com- plexity and preference. (Each person was asked to go over the scenes only with respect to a single rating.)
The results of this second effort (R. Kaplan, 1975*) were interesting in several respects. For preference and Com- plexity, the ratings that were repeated in the second study, the agreement between the original ratings and the new set was neither dismal nor encouraging. (In both cases, the correlation was .62.) Since participants in the original study had rated scenes in terms of both variables, these results suggested that ratings of preference may, in fact, have biased the Complexity reactions. This conclusion was reinforced by looking at some of the other findings. The correlation between the independently derived ratings of Complexity and preference was —.47, quite a contrast to the positive .37 when the ratings were done by the same individuals.
Thus we seem to have traded one perplexity for another.
60
It is now apparent that the positive relationship between prediction of Complexity and preference was an artifact; but why the preference negative relationship when the ratings were made inde- pendently? A clue as to what was going on here was pro- vided by looking at the relationship between Complexity and preference separately for each of the two content domains that had emerged from the CIM approach. Pref- erence for the Nature category was again far greater than for the Urban, as had been true in the original study [t = 4.22 and 6.76, df = 32, p < .005, respectively). Complexity, however, was far greater for the Urban than for the Nature scenes. (This was also true in the original study but not to the same extent: t = 7.08 and 3.13.) Using independent ratings, the correlation between Complexity and prefer- ence was not significant within either of the two content domains. Thus the overall negative correlation reflected content, the fact that the less liked urban scenes were more complex (Fig. 2.6).
What of the "new" predictors? Based on all 56 scenes, the correlation between preference ratings and Coherence was .34 and for Mystery it was .50. Ratings of Coherence showed no difference for the two content domains, but the
61
the preference Nature scenes were rated significantly higher on Mystery for nature (t = 4.09). Among the 22 Nature scenes, preference was
greater for scenes that had been rated as having more Mys- tery (and, to a lesser degree, this was also true for Coher- ence).
Although looking at the correlations is an intuitively satisfying approach to these data, a more appropriate pro- cedure involves regression analysis. Such analyses look at the combined effectiveness of a series of predictors in ex- plaining preference, summarized in a statistic called R2. Part of the effectiveness of any particular indicator might be attributable to its joint effect with another indicator (to the extent that the predictors themselves may be in- terrelated). Regression analysis excludes this "double counting." Thus the effectiveness of each predictor is com- puted by excluding the potential effects of all the others in- cluded in the equation.
Using regression analysis, the results of this initial study (using independent ratings of the three informational fac- tors) were very promising. R2 was .49, indicating that the three factors together accounted for almost half the var- iance. Each of the three factors was significant, with Mys- tery the most powerful (partial r = .56), Coherence a some- what weaker predictor (partial r = .33), and Complexity a negative factor (—.39). For the 22 Nature scenes taken by themselves, only Mystery and Coherence were significant predictors (R2 = .42), whereas for the 12 Urban scenes by themselves none of the predictors was significant.
Later Studies
Several other studies have included ratings of infor- mational factors, in addition to preference reactions. In most instances the ratings of the informational factors have been performed by a small group (four to eight in- dividuals) who rate the set of scenes sequentially (often in different orders and over several sessions), each time con- sidering a different factor. The intention is to have the panel members fully understand the meaning and nuances of the informational factors. In other cases, notably Her- zog's studies (1984*, 1985*, 1987*, 1989*) and Herzog and Smith (1988), the predictors are rated by groups of students who are given no more than a simple explanation of the in-
62
formational factor. Typically, in Herzog's studies any one prediction of participant is asked to rate the scenes in terms of a single preference factor or at most two of them.
The choice of predictors has also varied across studies. Mystery, Coherence, and Complexity have been included most frequently. Legibility was added to the Preference Matrix more recently and has thus received far less sys- tematic attention. Other variables have also been included, notably Identifiability (Herzog, 1984*, 1985*, 1987*), Spa- ciousness (Anderson, 1978*; Gallagher, 1977*; Herzog, 1989*), Smoothness, Openness, and Ease of Locomotion (R. Kaplan et al., in press*), and variables drawn from Ap- pleton's (1975) Prospect/Refuge model (Herzog, 1989*; Woodcock, 1982*). Though the predictors may have been called by the same name, the definitions that have been used have varied. Woodcock's use of Legibility, for exam- ple, comes much closer to Coherence as used in other studies.
The approaches to data analysis taken in the different studies have also varied widely. Herzog has generally ex- amined the role of the predictors in the context of specific perceptual categories, where they can help interpret the basis for the category but are limited by the small number of scenes making up a category.4 Other studies have generally reported either correlations or significant regres- sion results, sometimes without indication of the direction of prediction.
The discussion here draws on a relatively small set of studies. Included are instances for which we have access to the original data and could, therefore, perform compar- able analyses. The studies have in common that independ- ent ratings were obtained on a series of informational fac- tors. (Excluded are studies where the same participants were asked to rate the scenes in terms of their preference as well as any informational factors.) The studies include three doctoral dissertations (Anderson, 1978*; Gallagher, 1977*; and Woodcock, 1982*) and a master's thesis (Ells- worth, 1982*). In addition, five of Herzog's studies (Her- zog, 1984*, 1985*, 1987*, 1989*, and with Kaplan & Kap- lan, 1982*) and the R. Kaplan et al. (in press*) study are included in these analyses.
One would expect the informational predictors to be af- fected by the environmental context (or contexts) included in any particular study. The studies that are reviewed here varied widely in this respect. The studies by Gallagher,
63
the preference Herzog (1989*) and Herzog et al. each included a mixture for nature of built and naturai settings. In Gallagher's case, the study
area is a corporate headquarters (with the building visible in some scenes) and its surrounding lawn and landscaped areas. In the other two instances, the focus is on diverse urban places. In two studies, Ellsworth and Herzog (1985*), only waterscapes are included. The scenes in both the Anderson and Herzog (1984*) studies are of woods and fields; the scenes in the R. Kaplan et al. study include some of these settings, as well as agricultural land, in a more rural context. Unlike these other studies, which draw on a regional context, the scenes Woodcock used represent three different biomes and were taken in many places around the world.
It would also be reasonable to expect preferences to be affected by many issues other than those included in the Preference Matrix. In chapter 3, for example, we will con- sider the role of familiarity in preference. The studies under review here vary widely with respect to how much experience the viewers have had with the settings they are rating for preference. In Gallagher's study the ratings in- volve a local setting, with views that some of the employee participants had from their windows. Similarly, Ander- son's participants were viewing scenes of the familiar local setting. In Woodcock's study, by contrast, it is quite unlikely that any of the participants had much experience with rain forests.
Given this assortment of studies and variations in all matters that may be pertinent to the question, can one say anything at all about the prediction of preference? Table 2.3 provides some interesting outcomes. It presents the results of regression analyses, showing the R2 value as well as the partial correlation coefficient if it was significant (at p < .05). Results from the replication of the original study, discussed in the previous section, are included in the table as well. In all but the four Herzog studies (1984*, 1985*, 1987*, and 1989*), the regression analyses used only the in- formational factors in the Preference Matrix (Complexity, Coherence, Mystery, and Legibility), and these only if two or more were part of the study. In the case of the four Her- zog studies the regression analyses also included other pre- dictor variables, as indicated in the summaries of the studies in Appendix B, and this may partially account for the generally higher R2 values indicated for these studies.
64
I I I
X X X X X X X X
I I
X X X X
CO rl rl O)
CO CO CO CO
^ ^* ifl 03 CO CD CO CO 't CO lO
I I I
rl CD O) o n rl n W N CO CO W N CO
I I I I I I I I
CO 05 CO CO
rl 10 O lO o O O O
rl rl rl O O lO o
o o o o
o o o o o o o o
lO'jiiniOrlONNCO C0tqisNrlN05<t^
^dcoco'dcocdsd
rl ^ rl rl
OOJ^ONCOCDSincoO) ririNW^rl^CO^Wrf
CM0500CSJOOOCO
CD
o £ t;
a a o
§ w s
CD
DO ^ O
t> m O)
co eg co co
CT> CD CT3
rl £j -g rl rl g
00 00 00 OJ
o o o
N N N £
U *4 U ^
CO O QJ C
00 N * Pi S 05 CO
£ * u ffi o X X X <
O CD O LO rl
CO
s I
? §
N ^3
f-H ft
X &
^5
CO +«,
.S 9
T3 .a
•B HP
.a co
o o
65
Generalizing across these studies, the table suggests the following:
1. In each of the studies the combination of these informational predictors yielded significant results.
2. Complexity was a significant positive predictor in only a single study (and a negative predictor, as discussed earlier, in the initial study, where the Urban scenes were less preferred and rated higher in Complexity).
3. Legibility's role is hard to judge. In four of the five studies where it was included, Legibility did not play a significant role. In Anderson's study it was found to be a negative predictor.5
4. Coherence proved to be a significant predictor in the majority of the studies where it was included; in one case it was the only significant predictor in the regression analysis.
5. Finally, Mystery is the most consistent of the informational factors. (In the one study where it was not significant, the cor- relation between Mystery and preference [.45] was higher than the correlation between preference and either of the other two factors included in the study.)
Summary
The purpose of the Preference Matrix is to inform intui- tion. It is a framework, a structure for analysis. It suggests that the needs for understanding and for exploration are both important; one cannot replace the other. Similarly, the desire for both the immediate and the more inferential coexist. The results of the various regression analyses would suggest that on the understanding side, the im- mediately available information is particularly salient to preference. One wants to gain a quick grasp of the setting, and this is more readily achieved when it is coherent or well organized. As far as exploration is concerned, the em- phasis seems to be on the promise of information, rather than on the information that is immediately available. Preference is greatly enhanced when the scene suggests that more could be learned from entering deeper into the setting.
It is important, however, to realize that these results can- not speak to the full intention of the framework. Regres- sion analyses are useful for isolating the effectiveness of each factor in its "purest" form (i.e., having statistically eliminated potential joint effects with other factors). The Preference Matrix, however, is committed to the simul-
THE PREFERENCE FOR NATURE
66
taneous necessity of several factors, not the optimization prediction of of one of them. It would be most unfortunate to conclude preference from the regression analyses that the maximization of Mystery and the neglect of the other aspects are sufficient in designing a highly preferred setting.
We also want to emphasize that the framework pro- posed here is a conceptual tool that is guided by new infor- mation and insights. It is a model that continues to undergo change. Legibility in particular requires further develop- ment. Its basic intent seems solid: Environments that af- ford a good setting for functioning in a competent, comfor- table, and safe fashion should be more preferred. How this conceptualization manifests itself in diverse settings re- mains a challenge for further research.
PREFERENCE AND THE NATURAL ENVIRONMENT
People, like trees and snowflakes, differ from each other. They see the world through (conceptually) different eyes and bring diverse backgrounds to any new experience. People vary as well in what they like and dislike. Given this wondrous diversity, it is just as wondrous to find some strong and pervasive consistencies in the way people inter- pret the environment and in their preferences.
The strong preference for the natural world is a basic theme of this book. We see evidence of it in this chapter in the higher preferences for scenes that reflect nature as op- posed to more human-influenced elements. The preference for and importance of trees are evident in the results dis- cussed in this chapter and will receive further corrobora- tion in chapter 5. The sensitivity to natural elements and the preference for them have now been substantiated in numerous studies (see Schroeder, 1988, for an excellent review).
That is not to say, however, that all that is natural is pre- ferred. The challenge, then, is to determine what differen- tiates the natural areas that are more favored from those that are not. As we have seen, examination of most and least preferred scenes in any particular study provides some insights into this. Analysis of the relative prefer- ences for empirically determined perceptual categories also sheds light on this issue. Finally, the studies that sys- tematically assess attributes that are potential candidates
67
the preference for "preference predictors" offer yet another way to ap-
for nature proach this concern.
From these very different forms of analysis come some very consistent conclusions. Preference seems to be in- timately related to effective functioning. This relationship is generally not a result of conscious calculation. In fact, people are often willing to indicate their preferences based on the briefest glimpse, permitting no opportunity for care- ful reflection. Yet, ratings based on the briefest durations seem to be similar to those made in a less hasty fashion.6 Of course, such preference judgments are made by all of us very frequently, and not only in a research context. That one should favor a path - literally and figuratively - that facilitates functioning is hardly surprising.
The next step in the quest, then, is to identify some of the properties that characterize effective functioning. In an environment that fosters effective functioning, one might expect the individual to experience a sense of both safety and competence. One might also expect that individual to feel reasonably comfortable about the situation. These three aspects of an individual's feelings about an environ- ment are admittedly far from pure; they undoubtedly over- lap considerably. Nonetheless, they provide a useful, if rough-and-ready, indicator of the impacts of different sorts of environments. If the environment can make these more likely, one would expect it to be preferred. It is thus in- teresting to examine the qualities of preferred natural set- tings in terms of their relationship to these aspects of functioning.
We have seen that across very different environmental settings, scenes that are extremely open and lack any ele- ments that offer help in differentiating the depicted space are relatively lower in preference. Although such scenes might contribute to a sense of safety because it is easy to see in all directions, they offer little protection should one need it. Competence is undermined by the lack of land- marks and other features that help one maintain orienta- tion. Finally, the prospects for continued competence are likely to be reduced, given the limited opportunity for exploration.
Preference is also consistently low in settings that are blocked. A dense tangle of understory vegetation dominat- ing the foreground of a scene, for example, makes it un- likely that one could move readily through the area, thus reducing one's competence. The sense of lurking danger,
68
given the obstructed view, would also make the setting seem less safe and less comfortable.
Relatively uniform and short ground textures enhance the ability to locomote. Being able to move readily through an area gives one a greater sense of security and com- petence. Even if one cannot move easily through a setting, the opportunity to see through the trees can serve some of these needs. If there is visual access, the assessment of one's ability to function is also likely to be enhanced. It is far easier to avoid being surprised in an open, transparent forest than in a dark and dense one.
Coherent, orderly settings are likely to increase one's sense of competence because they are readily graspable. One can more readily determine where to focus attention and to sense what might lie ahead. Similarly, settings that are legible provide cues that help in maintaining orienta- tion, thus making safety, competence, and comfort more likely. The lack of Complexity suggests a setting that has too little to structure it, as is true of the wide-open areas. In a setting that has Mystery, one can learn more in a cognitively comfortable and safe fashion.
From these brief descriptions, it becomes apparent why the informational factors in the Preference Matrix help explain characteristics of preferred environments. The spatial definition or structure of an area, the textures that help one decide about the ease of locomotion or visual access, and the invitation to enter the scene to learn what cannot be determined from one's present vantage point are all powerful yet subtle qualities of the environment. With- out realizing it, humans interpret the environment in terms of their needs and prefer settings in which they are likely to function more effectively.
NOTES
1. Habitat selection is the technical name given to the inclina- tion to prefer environments that make successful adaptation more likely. Among vertebrates, habitat selection is a wide- spread tendency (Woodcock, 1982*). What this means is that animals tend to show a preference for the kind of environ- ments in which their species prospers. In some instances this occurs even if the animals have been raised in the laboratory and have had no direct prior experience with the environment in question (Wecker, 1964).
2. The original version of the Preference Matrix was included in
69
THE PREFERENCE s. Kaplan and Wendt (1972) and subsequently modified (S.
FOR NATURE Kaplan, 1975, 1979b). In those papers, Making Sense and In-
volvement were the labels for the informational domain. The use of Understanding and Exploration for these concepts began with S. Kaplan (1987a).
3. A vivid demonstration of the relationship between a failure to understand and an emotional reaction can be found in the classical study of Bruner and Postman (1949). In this fairly typical exploration of the human perceptual process, an image was flashed on a screen and the task was to identify it. If the image was not identified at the initial duration, say 1/100 of a second, then it was projected for a slightly longer duration, and so on until it was correctly identified. In this particular study the images were playing cards; to make the task a bit more challenging, they were printed in the opposite color from the usual pattern. In other words, spades were red and hearts were black. The performance of one particular in- dividual on this task was especially striking. The duration had to be increased steadily because of the participant's in- ability to identify the pattern. Finally, the image appeared on the screen for 0.3 sec, quite a long time for a study of this kind. The response to this extended duration was as follows: "I can't make the suit out, whatever it is. It didn't even look like a card that time; I don't know what color it is now or whether it's a spade or a heart. I'm not even sure now what a spade looks like! My God!" One must admit that this is a rather trivial situation. Nothing terrible is going to happen if the pattern is not correctly identified. Yet the concern to com- prehend, to make sense of the world around one, is urgent no matter what the situation.
4. Herzog has provided us information to include in the anal- yses here that, in some cases, does not appear in the pub- lished work. In other instances too, the material included in this book, though referring to published works, includes analyses that were carried out for the present purposes.
5. There are theoretical grounds for considering Identifiability as a component of Legibility. Herzog has explored this pre- dictor in several studies. In the 1985* study, Identifiability was not a significant predictor, whereas in both the 1984* and 1987* studies it was. In the Herzog, Kaplan, and Kaplan (1982*) study, an interaction between Identifiability and the content of the scenes led to a significant negative prediction. The scenes with strongest Identifiability, the Alley/Factory category, were also the least liked.
6. Comparison of preference ratings when the scenes are pre- sented for 15 sec as opposed to very brief durations (e.g., 10 msec) have not been completely consistent. In some cases, preferences have been more extreme (disliked categories receiving even lower ratings and like categories receiving higher ratings) when the scenes were seen only briefly (Her-
70
zog, 1987*; R. Kaplan, 1975*). In other cases,the opposite pat- PREDICTION OF tern was found (Herzog, 1985*). In still other cases, the means preference were either higher (Herzog, 1989*; Herzog, Kaplan, & Kap- lan, 1982*) or lower (Herzog, 1984*) at the short durations. Whether these preferences are statistically significant or not, a more important point is that even at the very shortest durations individuals have no difficulty in making these judgments, and they are substantially no different in their preference than when the scene is available for a prolonged time. Given that judgments of the environment are often made based on no more than a glimpse of the passing scene, this pattern of findings is not really surprising.
71
CHAPTER 3
VARIATIONS: GROUP DIFFERENCES
Several decades ago Florence Kluckhohn (1953) sug- gested certain "basic human problems" that must be addressed by all cultures. Her thesis was that each culture develops a dominant solution to each of these con- cerns but that the solutions also show identifiable varia- tions for specific groups within the culture. One of the themes Kluckhohn included in her short list of basic human problems was "man's relation to nature." All cultures must address this question, but within any culture there will be dominant and variant values attached to it.
Inclusion of this theme as a basic human problem is in it- self noteworthy. Kluckhohn's interest in this issue was mostly from the perspective of the balance or domination of humans and nature. From the material presented in the previous two chapters it would seem that, taken more generally, the role of nature is, indeed, a basic human thems. Chapter 1, on categorization, provided some in- teresting support for the relevance of the balance between the natural and the built in the way the environment is experienced.
The material in chapter 2, on preference, documented the consistent findings of strong preference for natural set- tings. In fact, the literature in environmental preference is rich with examples of consistency. Studies using diverse environmental contexts, varying ways of presenting infor- mation about the environments, and groups differing in various background characteristics have repeatedly re- ported high levels of agreement in preference ratings. In certain respects these similarities may be even more im- portant than the differences that are the focus of this chap- ter. The fact that such consistencies have been found across many studies supports the notion of dominant values with respect to the role of nature.
72
Our attention in this chapter turns to the variations on variations: the theme of preference. That there should be variation in group differences preference must come as no surprise. In fact, preferences are often thought to be completely idiosyncratic - "there is no accounting for taste." Complete idiosyncrasy, how- ever, has hardly been evident in the research on environ- mental preference. Rather, the pattern is one of strong con- tinuities and similarities. At the same time, however, there are some important differences.
The major factor accounting for differences is familiari- ty. Familiarity is the product of experience, and experi- ence comes in many forms. One gains such familiarity from many circumstances, such as where one lives, where one has visited, what one has studied, and the cultural norms of one's group. Some of the ways in which these cir- cumstances affect preferences are explored in this chap- ter. In examining these findings themes from the previous chapters appear again. Thus content and spatial configura- tion are still dominant issues; understanding and explora- tion also still provide a useful framework. The relative importance or weighting of these different aspects of pref- erence, however, changes.
Though familiarity is a major factor in understanding these variations, familiarity is by no means a simple pre- dictor of preference. The notion that familiarity breeds contempt has truth value. So does the notion that one pre- fers the things one knows. It is not the case, therefore, that simply knowing that a particular group has greater famil- iarity with some environmental settings will be sufficient basis for understanding differences in preferences. On the other hand, recognition of differences among groups and differences between experts and nonexperts is vital if one is to design and maintain natural settings for diverse users.
The chapter is divided into three main sections reflecting different sources of variation in preferences. The first ex- plores familiarity or experience that is related to the geographic circumstances of residence and to the effect of direct exposure to an environment. The second section looks at cultural and ethnic variation, including the ques- tion of age and other bases for belonging to a "subculture." The third section examines the effects of formal knowl- edge and expertise on environmental preference and perception.
There are two disadvantages to this organization of the material. The first is that the three categories overlap; the
73
the preference second is that a particular study might be pertinent to more for nature than one of the three categories. In most cases, studies that
are discussed repeatedly are included in Appendix B (and are designated by * in the text) so that a more coherent un- derstanding of the study can be obtained in this fashion. Despite these disadvantages, however, we hope this organ- ization might make an intricate and potentially confusing body of material more meaningful.
FAMILIARITY WITH THE LOCAL ENVIRONMENT
It would be reasonable to expect that the experience of rural living would affect preferences for natural settings differently than urban or suburban residential patterns. The view of nature is distinctly different in each of these contexts, and the people seeking one or another of these settings may have different nature preferences to begin with. Despite these considerations, however, studies that have explored this background issue have generally found no relationship that is clearly attributable to the urban/ rural experience of the participants.1
Such findings may be surprising, but there are many reasonable explanations. Where one has lived for periods of time may provide a poor indication of familiarity with other kinds of settings. One also may not live in the kind of setting one would choose if one had such luxury. Further- more, the very categorization of rural, urban, and subur- ban may not usefully parallel the way people experience different kinds of natural settings. Such groupings of the environment may be appropriate from a land-use perspec- tive but may be far less useful as typologies of natural settings.
Where One Lives
By contrast, some consistent findings suggest that people relate differently to settings with which they have direct experience. The environment near one's home, for exam- ple, holds some special significance that is reflected in preference judgments. This is an area of great familiarity; it also is related to a vast array of issues that matter to the individual. The Swift Run Drain Study (R. Kaplan, 1977a*) provides
74
a direct analysis of the relationship between preference variations: and place of residence. The 5-mile stretch of this storm group differences water drain includes a variety of residential areas. The character of the waterway itself ranges from a relatively unmanaged, "wild" appearance to regions where mowed lawn reaches to the edge of the water in relatively kempt fashion to areas where the water is fenced off and the adja- cent land neglected (Fig. 3.1). Along one of its six distinct regions the drain is buried underground, the only evidence of its existence a row of mature willow trees in a flood plain that serves as backyards for rented dwellings. The photo-questionnaire used in this study sampled from each of these portions of the drain and also included scenes taken elsewhere to represent other potential problems and treatments of a storm drain system.
For each of the 32 photographs, participants indicated both their preference and how similar the scene is to their \
view of the waterway. Preferences for these scenes varied widely across the different segments of the drain. Pre- ferences were not, however, consistently greater for the more similar scenes. Where the scenes were low in prefer- ence, they were equally low or even lower in areas where similarity was high (Fig. 3.2). In other words, an undesir- able situation is not made less distasteful by its familiarity.
In other cases the average preference rating was clearly highest in the area where the view was most familiar. Four scenes showed extreme variation in preference, a range of 2.3 or more (using a 5-point rating scale) across the six regions. Three of these (Fig. 3.3) depicted grassy areas with willow trees and no visible drain. These views were highly preferred by the residents living along the stretch of Swift
Figure 3.2 These scenes were rated low in preference whether or not participants were familiar with them. The scenes in the top row were, in fact, characteristic of the drain where some participants lived. The lower left scene, by contrast, was taken in another city and was thus unfamiliar to everyone. (Equally unfamiliar, and taken in another city, was the scene that was most preferred by the study participants.)
76
Figure 3.3 Scenes where the drain is underground. These were far more preferred by the adjacent households than by other study par- ticipants.
Run where it is buried. Their ratings for these scenes were more than a full scale point greater than the ratings at any other region; across these other regions the variation in preference was also quite large. The fourth scene (Fig. 3.1, upper left) had a mean rating of 4.6 by those living nearest it. Residents of other regions rated it between 2.0 and 3.1. We shall return to this scene later in the chapter.
The relationship between preference and familiarity with the nearby-natural environment was also examined in the photo-questionnaire study of multiple-family hous- ing (R. Kaplan, 1985a*). Here residents indicated the availability of a setting such as the one pictured as well as their preference for it. Two of the perceptual categories derived from the preference ratings consisted of natural settings (discussed in chapter 1; see Fig. 1.3). Such natural areas were most preferred and low in availability. The least preferred categories showed buildings dominating
77
the preference the landscape (Fig. 1.2, top). These were midrange in for nature availability. There were two additional categories, both
midrange in preference. One of these was rated as within walking distance by most participants (top row, Fig. 3.4), whereas the other was as unavailable as the nature scenes (bottom row, Fig. 3.4). Figure 3.5 summarizes these find- ings. Certainly, this pattern of findings suggests that preferences are not a simple outcome of what one can ex- perience most readily.
Participants in this study also indicated how readily they could see certain features from their residence and how satisfying they found these views. Here again we find a complex pattern (Fig. 3.6). Views that are highly familiar (parking areas, large mowed expanses, trees, and land- scaping) vary greatly in preference. At the same time, views that are highly preferred (woods, gardens, large trees, and landscaping) may not be very available. Absence
78
Parkland • Nature
VARIATIONS: GROUP DIFFERENCES
Landscaped
• Open, Residential
Building — dominated • Anonymous
1.5 2.0 2.5 (very close = 3)
Availability
Figure 3.5 Relationship between Availability and Preference for photograph-based groupings.
• woods
• garden
• field a park
larSe •small trees
trees • landscaped area
• large mowed area
children's play area
highway
•power line busy street
I
1.5
2.0 2.5 3.0 3.5
How readily seen from residence
parking area
4.0
Figure 3.6 Relationship between dominance of view and satisfaction with view, based on verbal items in questionnaire.
79
may indeed make the heart grow fonder, but presence need not mitigate appreciation.
Two other studies shed some light on the importance of place of residence in relation to preference. In both cases a comparison is afforded between place of residence and place of work. In these examples, familiarity may be relatively comparable, but other factors would seem to af- fect the judgments.
Gallagher's (1977*) study included both employees of a commercial facility and residents living near it. The 15 acres of grounds surrounding the facility, CUNA head- quarters in Madison, Wisconsin, had been redesigned, starting in 1972, to include landscaping that requires less direct management - a prairie restoration using natural landscaping principles. The photo-questionnaire consisted exclusively of views taken on the grounds of the facility, including areas that were kept as lawns, prairie areas with their tall grasses (Fig. 2.4), and regions to which some of the plant material originally included in the landscaping had been moved. These latter examples had relatively more trees (see Fig. 1.7).
Gallagher found the employees and neighbors differed significantly in their preference for the prairie scenes, with the neighbors showing greater appreciation. Among the neighbors, those living in the adjacent apartments were much more favorable toward the prairie restoration than were the nearby homeowners; the latter were significantly more favorable toward the views of lawns. In addition to the preference differences, Gallagher also reported strik- ing differences in the correlation between preference pre- dictors and preference ratings. Coherence was a negative predictor of preference for the apartment dwellers (—.44) whereas it was positive for the homeowners and em- ployees (.52 and .42). Similarly, the role of what Gallagher called "naturalness" (the unmanaged appearance) was op- posite for these groups (.49, —.56, and —.37). Mystery and the role of trees were equivalent in all cases.
Several factors may help explain the differences be- tween the two neighbor groups. Gallagher's sample of apartment dwellers was very small, and he is cautious in interpreting the results. However, these individuals were very favorable to the natural, unmanaged area. The home- owners, by contrast, remarked on the "messy" appear- ance. Their own residences tended to be landscaped tradi- tionally with large lawns and a specimen tree. It is not
BO
surprising that homeowners and renters would have dif- variations: ferent concerns for the appearance of the area immedi- group difference ately surrounding their residences. The fact that the apart- ments were relatively expensive may also play a role; the concern for order in the nearby-natural environment seems to be somewhat less salient at higher socioeconomic levels. Furthermore, the availability of nearby open space was a major reason for choosing these "luxury" apart- ments for many of their occupants.
The second study that permits comparison of place of residence and place of work involved public input in early design phases for a downtown vest-pocket park (S. Kaplan & Kaplan, 1989). Because the input was gained prior to construction, physical models were used for simulation, and these were kept relatively low in detail (Fig. 3.7). Par- ticipants were asked to rate their preference for each of 24 photographs of these simulated park environments. They
Figure 3.7 Photographs of simulations used in a citizen survey of a proposed small urban park (S. Kaplan & Kaplan, 1989). The models and these photographs are the work of Terry Brown and Charles Cares, the designers of the project.
81
the preference were also asked whether they lived and/or worked in an
for nature area near the proposed park.
The differences in preference between the people who work downtown and those who live there were substantial and consistent. Individuals who work in the area were much more positive in their ratings. For them the park, once completed, would offer a lunchtime respite and a pleasant oasis while running errands. For the nearby residents, however, the proposed park was viewed with some trepidation. Would undersirable individuals and "muggers" hide behind bushes and structures? The scenes were interpreted with such concerns in mind (and, in many cases, expressed). The park designers attempted to deal with these concerns in their plan for the park. They were apparently successful in doing this; some years later when the park was evaluated after it had been in use, the local residents and nearby workers were equivalent in their praise and did not differ in their judgments of the safety of the park (R. Kaplan, 1980).
Locals and Tourists
A similar contrast to the work-residence distinction can be made between local groups and occasional visitors. Here again there are likely to be differences in degree of famil- iarity, but other factors must also affect one's view of home range.
A study of scenic routes in Michigan's Upper Peninsula (R. Kaplan, 1977a*) used a photo album with each page showing four different views one might come upon while driving through the region. Participants were asked to evaluate the foursome in terms of preference. The sample included both year-round residents of the area and in- dividuals who happened to be visiting the region. Over all the scenes, the locals showed significantly lower prefer- ence than did the tourists. The locals, however, were more differentiated in their preferences (Fig. 3.8). They greatly preferred scenes of open forests to dense forests, whereas the visitors much preferred forests to flat farmland, with- out distinguishing the forest density in preferences.
Miller's (1984*) sample consisted of ferry passengers who were asked to indicate their preferences for views of the British Columbia coastline. One of the groupings that
82
Figure 3.8 Locals and visitors to the area differed in their prefer- ences for scenes such as these from Michigan's Upper Peninsula.
emerged from the Category-Identifying Methodology was Clearcuts and Natural Bald Spots. These scenes were significantly more preferred by passengers residing out- side British Columbia (visitors) than by passengers whose residence is more local. Out-of-province participants were less likely to recognize these scenes as showing timber har- vesting. The local residents, by contrast, showed greater appreciation of the Low Flat Shorelines grouping. Here again familiarity may play a role, with the tourists' expec- tation for shoreline topography negatively impacting their preferences when shown scenes lacking in relief.
These studies are interesting both for the conclusions one can draw and for the absence of regularities where one might have expected them. It is clear that the local pop- ulation is likely to have strong feelings about the local landscape. It is also clear that there are important areas of differences between locals and visitors. Attracting tourists
83
the preference may come at the cost of the affections of nearby residents.
for nature One conclusion that one cannot make is that the very
familiar environment is necessarily more preferred. The characteristic flat farmland scenes of the Upper Peninsula were more preferred by tourists than by locals, whereas in Miller's study the flat shoreline scenes were more pre- ferred by the provincial residents. Given the predictably strong concern of the locals, the potential for conflict be- tween local and visitor preferences, and the difficulty of predicting the specific pattern of preferences for any par- ticular locale, obtaining preference data at the local level would appear to be essential.2
Nature Hikers
A very direct way of exploring the issue of familiarity or experience is to study preferences both before and after a visit to a particular site. If individuals have never before been to the site, the question of familiarity is relatively well controlled and changes in preference can more easily be attributed to the experience.
As part of the Outdoor Challenge program (discussed more fully in chapter 4), participants were asked to rate 24 scenes in a photo-questionnaire (R. Kaplan, 1984b*). Two sets of ratings were printed beneath the photographs: "How familiar are you with scenes like the one shown in the picture?" And "How much do you like the scene?" All but three of these scenes reflected the various forested en- vironments encountered in the course of the program. In- cluded were relatively dense forests, open woods, swamps, and rocky areas. None of the scenes would be considered particularly scenic. The three "control" scenes, those not taken in the area of the wilderness trip, depicted common roadside scenery, showing relatively open coun- try with some trees and coarse-textured vegetation in the foreground.
These ratings were completed on two occasions: im- mediately prior to departure on the 9-day arduous wilder- ness trip and at its conclusion. The results are based on 49 participants who went on the trip in eight different groups in the course of two summers. Of these, 22 were adult par- ticipants (17 women and 5 men, ages 19-48) and 12 boys and 15 girls who were high school students.
Familiarity ratings were significantly greater at trip's
84
end than before departure for all but three scenes. These variations:
were the three scenes that were not taken in the area of the group differences
wilderness trip. Thus changes in the familiarity ratings
provide a strong indication that particpants were aware of
the kinds of habitats and ecosystems through which they
traveled.
With the exception of four scenes, preference ratings did not change significantly from the initial to the final rat- ings. The only scenes that showed significant preference changes were the four comprising the Swamp grouping (based on CIM). For these, preference declined (from 3.6 to 3.1, on a 5-point scale). Here much experience with tra- versing swampy areas seemed to have bred a certain amount of distaste for such habitats.
Both Hammitt (1978*, 1987) and his student Keyes (1984) have studied the relationship between familiarity and pref- erence with respect to a short hiking experience. The brevity of the hike presents the problem of asking in- dividuals to rate the same scenes in a matter of a few hours. The various studies in this series have approached this issue in a variety of ways, permitting comparison of ratings before and after the hike by different individuals, comparison of preference and familiarity ratings by same or different individuals, and comparison of ratings by same individuals using partially overlapping sets of photo- graphs. Each of the studies has been noteworthy for the large sample sizes.
Several findings have been consistent across these stu- dies. (1) Even the single hike along a nature trail (in Cran- berry Glades, West Virginia, or the Trillium Gap Trail in the Great Smokies) leads to extremely high familiarity ratings. Individuals are clearly aware of the scenes along the trail and recognize them in the photographs. (2) Both preference and familiarity ratings by different groups show remarkable consistency. (3) The relationship be- tween preference and familiarity is generally positive, but there are also very memorable scenes that are not pre- ferred. A scene of an uprooted tree, for example, receives low ratings on preference but high ratings on familiarity.
Summary
Direct experience and knowledge of a place can clearly af- fect preference. It is less clear, or predictable, what the ef-
85
feet will be. Visitors very rapidly become familiar with an area and can recognize scenes of it. Nonetheless, the ex- periences of longtime residents of an area reflect both greater differentiation of the landscape features and dif- ferent appreciation of the common aspects of the land- scape when compared to visitors to an area. Furthermore, residents are likely to be attached to the nearby-natural en- vironment in ways that are quite different from more tran- sient users, such as people who work in the area.
Based on these aspects of familiarity or experience, however, it is difficult to say how preference is affected. If anything, the pattern of these results simply suggests that before changing the nearby-natural setting one might bet- ter determine what the locals like and dislike about it. It would be difficult for a planner or designer to foretell the consequences of change given the results reviewed thus far.
CULTURE, SUBCULTURE, AND ETHNICITY
Another way to think about "where one lives" is at a cultural, as opposed to individual, level. People who share a system of thought or perhaps a language (or dialect) and pass these along from generation to generation would pre- sumably experience the environment similarly and might have some common preferences. In addition, people who share an environment would presumably have greater familiarity with its vegetation patterns and seasonal ef- fects, which would be expected to influence their percep- tions and preferences.
For purposes of our discussion it is not important to draw clear lines between culture, subculture, and ethni- city. It is useful, however, to explore some of the shared preferences and perceptions of such groups. Studies have touched on these sources of variation, but the results must be viewed with caution. These studies have generally not been carried out in a way that would give one great con- fidence in generalizing widely. On the other hand, it is likely that insights about subcultural influences will be forthcoming only through the accumulation of such small- scale studies. Consistency across these efforts is well worth heeding.
THE PREFERENCE FOR NATURE
86
Cross-cultural Comparison
VARIATIONS: GROUP DIFFERENCES
The literature on cross-cultural comparison in preference for natural environments shows relatively high agreement when cultures are similar (Zube, 1984). Zube and his students have reported correlations between .76 and .89 among two Australian samples and American landscape architecture students, using photographs of Australian landscapes (Zube & Mills, 1976), and similarly high cor- relations in comparisons among Yugoslavian students, Italian-Americans, and various American groups, using scenes of northeastern American landscapes (Zube & Pitt, 1981). Others have reported consistent preferences when comparing Americans with Scots (Shafer & Tooby, 1973) or Swedes (Ulrich, 1983). These studies, however, have focused largely on correlational results.
In some of our recent work in this area we have found that high correlations between groups tell only part of the story. Though the groups may agree in their relative likes and dislikes of certain scenes (reflected in high cor- relations), they may nonetheless differ substantially in how much they actually like the scenes (mean rating). Furthermore, despite high correlations, it is possible that the groups differ in how they experience the environment, as reflected in perceptual categorizations (CIM-based re- sults). Such comparisons across different analytic methods are not available for the studies reported in the previous paragraph. Let us turn, then, to three studies that permit such exploration. Across these, the cultural influences that are examined entail both the landscape and the partici- pants. In Yang's (1988*) study, the focus is on oriental and western influences, whereas in the R. Kaplan and Herbert (1987*, 1988*) studies the comparison is between Western Australia and the north-central United States.
Yang's study is one of very few in the preference litera- ture that examines non-western landscapes; it is also un- usual for its samples. All 40 scenes in his photo-question- naire were taken in Korea, but they were selected to reflect Korean, Japanese, and western-style landscapes. (Korean- designed landscapes are characterized by fences, walls, and pillars, all constructed of quarried stone blocks. Rec- tilinear rather than curvilinear forms tend to dominate the highly structured settings.) For each style, distinct land- scape characteristics (water, vegetation, rock, and layout
87
Figure 3.9 Scenes in top row were among the most preferred in Yang's (1988*) study for both Korean and western samples. Both are waterscapes: Jeft, Japanese-style, right, Korean-style. Scene at bot- tom left is designated western-style and was among the most pre- ferred for the Korean sample; scene at bottom right is an example of Korean style and was among the six top scenes for the western sam- ple. Neither of these latter scenes is a waterscape, and each shows preference for a "nonnative" style.
of space) were included. Both Koreans (560 citizens and students) and western tourists (N = 110) served as parti- cipants.
Although the tourist sample's preference ratings were significantly higher (mean 3.3 as opposed to 2.8 for the Koreans), the relative preferences showed strong similar- ities. Of the six most preferred scenes for each group, four were the same (Fig. 3.9), and water was a dominant theme among the most liked scenes. The least preferred scenes
88
also showed overlap (three scenes common to the set of six variations: for each sample), and the use of quarried rock was the group differences dominant theme among the less preferred landscapes (Fig. 3.10). The samples were also similar both in their prefer- ence for Japanese landscape style and in their relatively lower appreciation for the style representing their own culture.
Figure 3.10 Top row: scenes that were among the six least preferred for both samples in Yang's (1988*) study. Bottom row: scene at left was among least preferred by Koreans though not by western sam- ple; scene at right was low in preference for the western sample but not for Koreans. All four scenes are examples of Korean style, and all include the quarried rock element. Scene at lower right provides a relatively unusual example of a water scene that received a low pref- erence rating.
89
the preference CIM procedures generated four categories for each sam- for nature p|6j an(j these showed both similarities and differences.
The most preferred category in each case reflected the preference for water and for Japanese style, with six com- mon scenes. In the case of the Korean-based category, ad- ditional scenes suggested that the common theme for that grouping was Water Surrounded by Vegetation. The scenes included only in the western-based grouping sug- gested naming the category Japanese-style Landscapes. The least preferred categories showed a similar pattern, with nine common scenes. The additional scenes for the Korean-based grouping more strongly reflected Rock with Sparse Vegetation, whereas the scenes unique to the western-based category were more uniform with respect to the style, and hence the grouping was named Korean- style Landscapes.
The other two categories for each sample were based to a large extent on the differences between western formal and western informal style, although not all the "infor- mal" scenes for one sample were, in fact, designated "western." For both groups the informal category was significantly preferred.
The strong cross-cultural similarities in Yang's study are surprising in some respects. One might certainly have ex- pected stronger preferences on the part of the Koreans for their own, dominant, style. On the other hand, the Korean sample's perceptual categories do not distinguish as clear- ly between the two oriental styles, as did the western sam- ple. The designation of "western" style might seem mis- leading to a western researcher, who would be likely to see distinct differences among these. The empirically gen- erated categories made it evident that the participants, regardless of cultural orientation, also did not perceive this as a distinct style.
The other two studies (R. Kaplan & Herbert, 1987*, 1988*) involved university students in the United States and in Western Australia.3 Though all the scenes were of nondramatic natural landscapes, in one instance the scenes were taken in Michigan and in the other Western Australian landscapes were used. First-year students at the University of Michigan and at the University of West- ern Australia are culturally distinct, though the cultures are relatively similar. On the other hand, the characteristic vegetation in the two instances is distinctly different.
The 55 slides taken in the rural western parts of Oakland
90
County, Michigan, included scenery typical of midwestern variations: and eastern portions of the United States, consisting of group differences abandoned fields, forests, and rural housing. There is little topographic variation in the region (Herbert, 1981*). The 60 slides of Western Australia sampled five landscape types found in the northern Jarrah forest: escarpment, up- lands, valleys, plantations, and eastern woodlands. The Australian and American scenes contrast in two important respects: the relative aridity of many of the Australian set- tings and the characteristic gray-green foliage of the domi- nant eucalypt species of the Australian landscape.
The results support previous findings when viewed in terms of correlations between the two samples. Correla- tion of .84 was obtained in each study. The mean ratings in the two studies show that each group favors its own landscape slightly (but significantly), with means of 3.3 and 3.1 across the respective slide sets. For the Australian landscapes there were four instances where the American students significantly preferred a scene. Two of these were of vistas (one with a river dominating the scene) that could have been taken in many parts of the United States. The other two showed a palm tree in one instance and bare trees in the other. Since bare trees are more likely to reflect disease than season in the Western Australian context, this difference is understandable. The palm tree was pre- sumably more familiar to the American sample than the eucalypts visible in most of the scenes. These results sug- gest relatively high agreement between the two groups and that familiarity enhances preference to some degree.
The comparison between the two groups turns out to be more interesting, however. If we use the Category-Identi- fying Methodology (CIM) separately for each cultural group in each study, some differences in both perception and preference become evident.
For the American scenes (R. Kaplan & Herbert, 1988*), the groupings were very similar and the preferences equiv- alent. Table 3.1 shows that the four categories generated by the CIM procedure can be described using the same label even though the scenes comprising the categories are somewhat different using the American-based and the Australian-based ratings.
The major differences between the two analyses in- volved somewhat different interpretations of the Vegeta- tion and the Pastoral groupings. The Australian-based Vegetation grouping included 12 of the 17 scenes in the
91
-a c
CO
c
CO O
"C
CD
§ <
c o
CO — N
CO *
oo Sfi «
S M S t:
CO CD CO CD
CD
E B
<^ CO CD Q<
t-H
g a
CO CD CD CO
S CO
CD CJ
CD Eh <D 4h CD
CO
73
CO ^
CD
ca
n co CO 3
CO
13
G cd
CU
CO
ca
CO
G cd
XI CU CO
cO
rl N <t P3
O O r-l
00 ^ t-- 05 CO CO N N
rH CM CO ^
O CO N CO 3| t-J CO CO CO
n <t h co
rH CM
r-l CM ^ CO
CO N ^ CO CO t CD N
CO CO Csl CSJ
rH (M CO ^
CO CM CO CO CO CM
t> rH O CO
^ t> to o
rH rH rH
M O
O _H OJ fH
2 Ph .2 G
O .2 'ca G
tH CU
o t3
to "co CO CU
Ph P<
CD
cj
1$
CU t-l
CO CU CO CU
pC
G CU (-1 cd
Oh
.5
CO SH
CU
£ G
CU
o
92
American-based version. Dropped were five scenes that were distinctly less green than the rest, and two grassy scenes were added. Thus, though the two Vegetation groupings are very similar, the one based on the Australian sample has a stronger sense of lush, green landscapes and more forested settings (Fig. 3.11). The preferences were es- sentially identical.
The scenes comprising the Pastoral category were quite different when based on the ratings by the Australian as opposed to the American students. Thirteen of 15 of the scenes in the American-based category are included, but eight are added in the Australian version. With these ad- ditions, the grouping is more suggestive of open, rural land than of a pastoral setting (Fig. 3.12). Whether using the Pastoral or the Open Rural version of this grouping, the mean rating is distinctly lower for the Australians (2.6)
r
Figure 3.11 Differences between American-based and Australian- based Vegetation category (R. Kaplan & Herbert, 1988*). Scenes in the top row were common to both the American-based and the Aus- tralian-based CIM results. Those in the bottom row were unique to each of the analyses.
93
Figure 3.12 Differences between American-based and Australian- based Pastoral category (R. Kaplan & Herbert 1988*). Top row scenes were included in both instances. The bottom row, however, repre- sents the different perception of the Pastoral grouping to reflect a more Open Rural landscape pattern on the part of the Australian sample.
than for the Americans (3.1). Although these scenes are far preferred to the Residential category by the American students, for the Australians they are the least preferred of the four clusters.
The conclusion that Americans prefer their more famil- iar landscape is quite misleading given these findings. It seems that, although the Australians are not directly fa- miliar with any of the scenes, the genre of the open rural setting is one they recognize readily. This is probably the most common nearby nonurban landscape in the area near Perth, Western Australia. In this case, a relative sense of familiarity for settings that are not highly preferred seems to breed contempt. For the remaining scenes, the two groups are equivalent in preference. Whether this means that the relative novelty of lusher landscapes enhances the
94
Australian students' preferences cannot be determined variations:
from these results. group differences
For the Australian landscapes (R. Kaplan & Herbert, 1987*), the perceptual differences between the two sam- ples were more extensive. Using the Australian preference ratings, three categories emerged: Eucalypt Forests, often with strong vertical tree trunks and gray-green foliage (dis- tinctly preferred by the Australians); Australian Scrubland which is quite open and arid; and Pastoral, Grazed, and Relatively Manipulated spacious areas (Fig. 3.13). The relative preference for the three groupings was similar for the two student samples.
Using the American preference ratings yielded five categories. These exemplify the kinds of issues that have consistently emerged in the perceptual analyses that are summarized in chapter 1, on categorization. Openness and smoothness of ground texture are salient properties of the perceptual process. Similarly, vistas afford visual access on a larger scale. Finally, trees play an important role in perception of the natural environment. Thus the combina- tion of tree cover and ground texture distinguishes among several of the categories (Fig. 3.14). The Vista scenes were significantly preferred by the American students. Whether this is a cultural effect or contextual (given this set of scenes, the vistas provided a bigger picture in the context of less familiar settings) requires further research.
The results of these two studies suggest that familiarity with the local natural setting has an effect on perception. The effect of such familiarity on preference is, once again, mixed. Knowledge of the local scene does not assure in- creased preference, and the novelty of foreign places may enhance it. How much "culture" plays a role in these mat- ters is an open question.
Subcultures
The study of the Australian landscapes (R. Kaplan & Her- bert, 1987*) had an added dimension that challenges cul- tural interpretation of preferences. Another group of Western Australians participated in this study. These were members of a Wildflower Society, sharing an interest in botanical and conservation matters. In exchange for a presentation by Eugene Herbert, then with the Forest Department of Western Australia, they consented to view
95
Figure 3.13 Examples of each of the Australian-based categories of the Australian landscape (R. Kaplan & Herbert, 1987*). Top row, Eu- calypt Forests; middle row, Scrubland; bottom row, Pastoral, Grazed, and Relatively Manipulated settings.
the slides and rate them in terms of preference. Whereas the two student samples, just discussed, permit cross- cultural comparisons, the two Western Australian groups can be assumed to share a culture and to be acquainted with the local natural environment. Nonetheless, these two groups differ in age and education (with more diver-
ge
Figure 3.14 Examples of four of the five American-based categories of the Australian scenes. Top row, Jeft, Forest/Forest Vista category, right, Open Woodland/Field category. Bottom row, left, part of a category labeled Rough-textured, Arid, Wooded, right, Vista cate- gory. The remaining, unpictured, category was Open, Smooth Texture.
sity in both respects on the part of the Wildflower Society members). Perhaps more important than either age or for- mal schooling, the two Western Australian groups differ in their interest in the natural environment.
Once again the correlation was high, .81, between the two Australian groups, and the composite mean prefer- ences for all 60 slides were virtually identical. However, for 22 of the scenes the Wildflower Society members in- dicated significantly greater preference, and for 11 scenes the opposite was true. Thus, for over half the scenes, the two groups were far apart in their appreciation of the local countryside.
The most consistent factor in the differences in pref- erences between these two samples related to the issue of use of native plants. Pines and pine plantations have been the source of considerable local debate, with some vocal
97
the preference citizens resenting the introduction of exotic species. The for nature Wildflower Society members could be expected to be sen-
sitive to this issue and to prefer virgin forests and scenes reflecting more typical native vegetative associations (and, in fact, they preferred scenes in middle row, Fig. 3.13). The Australian students, by contrast, showed no disfavor of pines (Fig. 3.13, lower right) or of farmland or of other less pristine indigenous bush. These strong differences suggest that familiarity can take several forms. Greater knowledge and concern for the types of species are evident in these preference differences.
The Wildflower Society sample was not large enough to permit a separate CIM analysis, and it is thus impossible to determine whether their perceptual experience is notably different from the students'. The differences in preference are evident, however, when one compares mean ratings for the two Australian groups, using the Australian-based categories. These differences are apparent in the greater distaste for manipulated landscapes and the greater ap- preciation both for arid, open, coarse-textured views and for scenes of characteristic Western Australian forest.
Different conclusions about similarities and differences in both environmental perception and preference would have resulted if the study had included any two of the three samples. Though both the Australian and the American (student) samples represent "western culture," clear dif- ferences exist between the American and Western Aus- tralian landscapes. The fact that the two samples that share the same landscape show major differences in pref- erences has important implications for public participa- tion, for the management of land resources, and for our un- derstanding of subcultures.
Similar subcultural differences have been found in many other contexts as well. Daniel and Boster (1976), for example, included 26 different groups in their validation study of the Scenic Beauty Estimation method. They report generally high agreement with respect to scenic preferences, with the exception of the four groups rep- resenting range (or cattle-growing) interests. Porter's (1987) study of preferences for the agricultural countryside also included several discrete public groups, varying in their familiarity with farming and farm scenery. The dairy farmers in the study region, the northwestern part of Washington, showed the most extreme range of prefer-
98
ences, far preferring the open farmland scenes to any other variations:
group. GROUP DIFFERENCES
Ethnicity
There are hints in the literature of differences in environ- mental preference between black and white groups. Across quite a few studies in distinctly different settings, the pattern of these differences is quite consistent. There is great reluctance to attribute these differences to ethnicity, however. In fact, in several instances the results are re- ported in such a way that it is difficult to determine that ethnic differences are present. Schroeder (1983), for exam- ple, describes the groups as differing in "urban" back- ground, with no mention of ethnicity. Zube and Pitt (1981) report distinctly different preferences on the part of a large sample of Virgin Island residents (as compared to Yugo- slavian students and various American samples). That the Virgin Islanders are black is true (Pitt, personal com- munication, 1986), but not mentioned. Given that many of the studies that have found differences between blacks and whites are based on small samples and that it is dif- ficult to determine whether these differences are not also attributable to other factors (e.g., urbanity, age, affluence), it is important to be tentative in drawing conclusions. At the same time, however, it is important to acknowledge that such differences may exist and that this aspect of sub- cultures is no less likely to lead to preference differences than are the various others.
Washburne and Wall (1980) provide a useful analysis of these issues. They argue that different recreational pat- terns on the part of American blacks cannot be explained in terms of hardships and "culture of poverty." "The resurgence of interest in ethnic cultural roots . . . supports the existence of true subcultures" (p. 11), and a pluralistic view of American society has replaced the assumption of assimilation as the goal of all minority groups.
Anderson's study (1978*) included black and white high school students and local residents. These participants were not urban. They lived in a rural area where forestry is a major factor in the economy (though at the time of the study unemployment was very high). When he went to the area to take the photographs for his study he consulted
99
the preference some local individuals about where he might find some for nature particularly attractive scenes. At least one black resident
explained to him that "there weren't any anymore; the trees are all grown up now." The discomfort with the forests was evident from other comments as well. For ex- ample, when asked for advice about getting back into town a black merchant provided some lengthy instructions. When asked whether it would not be shorter to take a dif- ferent route he replied: "You don't want to go through the forest, do you?" (Anderson, personal communication, 1977).
These are admittedly anecdotal descriptions. Ander- son's findings, however, support these impressions. For both students and residents, the white participants signi- ficantly preferred scenes of dense forests. The white students were more favorable toward the heavily manipu- lated (clearcut) areas, though neither student group was fond of these, nor were the adults. Among the residents, blacks expressed much greater fondness for what Ander- son called "planned spaciousness," though these were highly preferred settings for all groups. Not only is the tree cover less dense in these scenes, there are also suggestions of human influence, such as roads and picnic tables. (As with previous studies, here again the very high cor- relations - between .81 and .91 - when comparing resi- dents and students, blacks and whites, would fail to reveal the important differences among these groups.)
The preference pattern Anderson found in a rural con- text is surprisingly similar to results we obtained in the urban context. Participants in this study (R. Kaplan & Talbot, 1988*) were 97 black residents of three low-to- moderate-income black neighborhoods in Detroit, Michi- gan. These participants were asked to sort 26 photographs into five piles corresponding to their preferences. The scenes depicted typical outdoor areas, including unmani- cured wooded areas, lakes and rivers, landscaped parks, picnic areas, and residential street scenes.
Figure 3.15 shows some of the most preferred photo- graphs in the study (means well over 4.0). These included parks or neighborhood scenes with both natural and built elements. Many had a sidewalk or built pathway as well as park equipment or a picnic shelter or a portion of a build- ing. The ground texture was smooth, and the overall im- pression was of a well-kept area. Some of the least pre- ferred scenes (means below 3.0) are shown in Figure 3.16.
100
Figure 3.15 Among the scenes highly preferred by the Detroit residents.
The bottom left scene may look familiar (see Fig. 3.1), as it had been part of the storm drain study (R. Kaplan, 1977a*) discussed earlier. These unfavored scenes depicted un- managed areas with heavy undergrowth. Tree density, weeds, and a scrubby appearance contributed to these low ratings.
These preferences were also reflected in the par- ticipants' perceptions. The CIM-based category Local Parks and Walks groups together scenes depicting paved walks as well as other built elements in the context of na- ture. The top two scenes in Figure 3.15 are included in this cluster. Both Woods and Path and Woods and Water were far less favored and were similar in including larger and more densely forested scenes (top left and top right in Fig. 3.16, respectively). Pathways and small openings are evi- dent in the former, and a lake or river is in the foreground in the latter grouping. Preference for these two categories was equivalent (providing another example where water is
101
Figure 3.16. Among the least preferred scenes in the Detroit study.
not highly preferred). These perceptual groupings are similar to results reported by Getz, Karo, and Kielbaso (1982), who also studied black Detroit residents. They found higher preferences for trees on neighborhood streets than for larger wooded areas.
After completion of the sorting task participants were asked to describe aspects of the outdoor areas in the photographs that they particularly liked or disliked. Or- derliness, safety, and visibility within an area were the leading concerns reflected in these comments. Trees, built features, and neatness were most mentioned as assets. Though trees are greatly appreciated, the ones in the least liked scene (bottom left, Fig. 3.16) were singled out as negative instances. They were interpreted as looking dead, and the scene as a whole was described as disorderly and cluttered. Heavy undergrowth was associated with the threat of physical danger.
102
The same scenes and task were used in a pretest prior to variations: the Detroit study. The sample was small, including 21 group differences white and 10 black residents of Ann Arbor, Michigan. The responses of the two black samples (Ann Arbor and De- troit) were quite similar, with only three scenes showing significantly different ratings. By contrast, the differences between the white participants and either of the black samples were substantial. Not only were most ratings significantly different, but the pattern of most and least preferred was reversed in many cases. In fact, the scenes in Figure 3.15 were all rated among the least liked (means be- tween 2.5 and 3.0) by the white participants, and the scenes in Figure 3.16 were all among their highly preferred (means between 3.5 and 4.0).
As Schroeder (1988) has pointed out, some have con- cluded that blacks may be less interested than white Americans in nature and the outdoors. The results of our research do not support this contention. In fact, the in- dications are clear and strong that natural settings have great importance to the black Detroit residents. The pat- tern of findings across various studies, however, suggests that human influence, neatness, and openness are far more vital to some groups than to others.4
Age as Subculture
Teens are often identified as a subculture. In certain re- spects this is just as appropriate for other age brackets as well. It is also just as true with respect to age as it was for other group comparisons that it is at best difficult to ascer- tain whether findings showing group differences can ac- curately be attributed to that group designation. For exam- ple, if an older group is found to prefer scenes with greater human influence (as was true in Miller's 1984* study), is this difference attributable to the age per se, or are other factors more helpful in understanding the results? Ex- perience with the effects of fluctuating economies might well have been substantially greater for individuals now in their fifties and older than was the case for their age-mates a century ago. It must go without saying that background variables must always be viewed with caution.
Balling and Falk (1982) reported intriguing findings with respect to developmental variation in environmental pref- erence. Their study included four scenes for each of five
103
the preference biomes (savanna, deciduous, coniferous, rain forest, and for nature desert). The participants spanned a wide age range, from
students in grades 3, 6, and 9 to college students, "adults," and "14 retired citizens." They describe this sample of 548 participants as "extremely mixed" in terms of other back- ground characteristics. The slides were shown twice (with the order of the task counterbalanced), once to obtain ratings of preference as a place to live and the other time as a place to visit.
Two aspects of the results are particularly noteworthy in the context of this discussion. First, for the two youngest groups (ages 8 and 11) the preference for savanna scenes was significantly greater than for deciduous and con- iferous settings, although the latter must be far more familiar to the children. For all the other age groups these three biomes were equivalent in preference. Desert ancL rain forest scenes were always significantly less liked than the other three habitats. Second, the preferences of the 15- year-olds were consistently lower than those of the adja- cent age groups. The plots of preferences for the various biomes all have a remarkable dip at that point in the curve.5 ,
Balling and Falk do not discuss this striking preference pattern of the adolescent participants. Zube, Pitt, and Evans (1983) also studied "lifespan developmental" var- iations in preferences and included younger age groups in their investigation. Unfortunately, they combine 12-18- year-olds into a single age group. This group is not distinct from the 19-35-year-old group in the correlational analy- ses presented. The report of the study also includes no in- formation about the magnitudes of preferences for the diverse scenes that had been used. It is thus impossible to tell whether the actual preferences for different environ- ments differed across the lifespan.
Medina's (1983*) study sheds some further light on this question of young people and preference. He found the seventh and eighth graders in his study (from predominant- ly black Detroit, Michigan, schools) to have strongly dif- ferent preferences than the adults (individuals engaged in environmental education all over the United States). The two groups thus differed in terms of at least four sub- cultural categories: age, ethnicity, place of residence, and knowledge.
What the students seemed to appreciate most were set- tings that suggested activity, places to do things. Scenes
104
with buses, for example, suggest that one can go some- variations: where, and scenes with stores suggest that one can buy group differences something. By contrast, tree-lined streets, preferred by the environmental educators, may be nice to look at but communicate little action. One can interpret the youths' preferences from various perspectives: (1) The preference pattern is similar to the black preferences discussed pre- viously (higher preference for more developed areas). (2) Like Balling and Falk's 15-year-olds, these youths show generally lower preference for more natural settings. (3) The pattern also resembles the findings that people prefer what is most familiar (greater preference for row houses and front-yard lawns than for places dominated by trees). (4) The results also fit a stereotype of the teen subculture (an eagerness to go places and do things, at least in the out- door context). Whatever the interpretation, the vast prefer- ence differences between these young teens and the people who are involved in teaching about the environment must be recognized if environmental education is to achieve its goals.
Summary
Although these studies range widely with respect to hu- man groups and environmental contexts, they show that the balance between nature and human influence is indeed an important factor in understanding preferences. It is easiest to see the subcultural differences as the variant solutions to this theme. For some, nature is favored if it is groomed, orderly, and with indication of human-sign. For others, nature that appears unmanaged and wild is the more preferred. To label only one end of such a suggested continuum as "nature" would undermine the strong posi- tive feelings common to most of these groups with respect to the natural environment.
FORMAL KNOWLEDGE OR EXPERTISE
To a large extent the experiences we have discussed so far have been vicarious or circumstantial. As a resident or visitor or member of a particular group one has encounters with the environment that may be affected by those cir- cumstances. By contrast, the comparisons examined in
105
the preference this section are related to a more intentional pursuit of
for nature knowledge about the environment.
Although this is a convenient distinction to draw, it is important to realize that it is only somewhat helpful. The major problem in making the distinction is that learning goes on all the time, and its effect on perception and pref- erence happens without intention or awareness. In chapter 1 we spent quite some time developing the concept of expertise and discussed how experts' perception is dif- ferent from people who do not share that expertise. In a sense, residents have an expertise about their local en- vironment, and members of a subculture have an expertise with respect to that group. The kinds of expertise we will include in this section, however, are related to professional training.
Preference and Knowledge
We know of relatively few results that point to differences in preference attributable to knowledge about the environ- ment.6 The wilderness data, showing a decline for prefer- ence of swamp scenes after firsthand experience with such settings, provide some indication. However, the knowl- edge of the swamps obtained from the outing was more physical than cognitive. The different appreciation of the environment exhibited by the Wildflower Society mem- bers may or may not be related to formal knowledge about the role of native as opposed to exotic species. There is a strong element of nationalism involved in this issue; one can also favor conservation in principle without exten- sive knowledge.
A study by Hodgson and Thayer (1980) addressed the possibility that preference is influenced by the suggested interpretation of the scene. Scenes labeled with more "natural" labels (lake, pond, stream bank, and forest growth) were ranked more favorably than when the identi- cal scenes were given labels suggesting greater human in- fluence (reservoir irrigation, road cut, and tree farm). Thus the information provided by the label became part of the interpretation of the scene. Presumably, the labels without any pictures would reflect similar differences in preference.
Gallagher's (1977*) study of prairie restoration at the CUNA facility in Madison, Wisconsin, included a question
106
about how informed the participant was about the restora- variations:
tion and its intent. For the small handful of people for group differences
whom he had such information, there was a suggestion of
greater preference for the prairie grouping (p < .10) on the
part of the informed participants.
Buhyoff et al. (1979) studied whether information about beetle damage affected aesthetic judgments. Of the 10 scenes in the study, 8 showed various stages of southern pine beetle crown damage, with colors varying from yellow-orange and red-brown to black, while the remain- ing two scenes taken along the same stretch of the Blue Ridge Parkway had no such damage (control slides). Four groups participated in the study: introductory forestry students, outdoor recreation students, a nonforestry group including Sierra Club members and residents, and a forest- ry professional group. For all but the last of these groups, half the participants were told that they would be assess- ing scenes with southern pine beetle damage, and the rest were told that they were participating in a "landscape preference test." (The professional group was assumed to know that the scenes reflected disease.) Scenes were pre- sented pairwise, and individuals were asked to pick the scene they least liked in each pair.
Of the two control slides, one was ranked most liked regardless of background and information. The other one was ranked second by the informed groups and third by the others. The relative preferences of the scenes showing beetle damage, however, were significantly different for the informed and uninformed groups. For the informed groups, the ranking of preference corresponded very close- ly to the degree of damaged vegetation in the scene, and there was virtually no difference among these groups, de- spite their different background knowledge. For the unin- formed groups, by contrast, preferences were actually en- hanced with increasing amounts of orange-brown foliage since they probably interpreted this as fall coloration rather than the result of infestation.
Keyes's thesis (1984) explored knowledge obtained by virtue of eight interpretive signs along a hiking trail in the Great Smokies. The signs each contained two to three sen- tences along with an illustrative sketch and referred to in- formation specific to the trail. The signs were mounted in such a way that it was possible to install them on some testing days, permitting direct comparison of preferences
107
the preference between hikers who had the signs available and those who
FOR NATURE did not.
Keyes reports that of the 298 hikers exposed to the signs 92% indicated that they read at least half of them and