Why do Preferences Differ between Scene Types?

ENVIRONMENT Purcell et al. / SCENE AND TYPES BEHAVIOR / January 2001

WHY DO PREFERENCES DIFFER BETWEEN SCENE TYPES?

TERRY PURCELL is a psychologist in the Department of Architectural and Design Science in the Faculty of Architecture at Sydney University, Australia. His research interests are related to the cognitive processes involved in the formation of mental representations of and the generation of affective experience about the environment, similarities and differences within and between cultures in environmental experience, and cognitive processes involved in design problem solving, particularly the role of sketching and drawing and their relationship to imagery and working memory. ERMINIELDA PERON is a psychologist in the Department of General Psychology at the University of Padua in Italy. Her research interests cover all aspects of environmental learning and memory, the development across the lifespan of mental representations of the environment, and the role that affect plays in environmental experience. RITA BERTO is a graduate in psychology from the University of Padua and is currently working on a number of environmental psychology research projects in the Department of General Psychology in that university.

ABSTRACT: Groups of subjects judged one example of two different types of outdoor scene on each of the items of the Perceived Restorative Scale, on two preference scales and a familiarity scale. It was argued that the previously demonstrated large variations in preference between different types of scenes were the result of participants using the restorative value of a scene as an implicit frame of reference for the preference judgment. Preference and the Perceived Restorative Scale score correlated .81, whereas familiarity and the Restorative Scale correlated .31, and preference and familiarity correlated .32. This result supports the hypothesis regarding the use of the restorative value of a scene as an implicit frame of reference for preference judgments. It is further argued that variations in the preference and restorative value of scenes may be associated with fractal geometry.

Beginning in the 1960s, research was published that addressed the question of people’s preferences for landscapes. Although the area is now associated with an extensive literature relating to different types of landscapes, different geographical contexts, and the possible physical attributes of the landscape ENVIRONMENT AND BEHAVIOR, Vol. 33 No. 1, January 2001 93-106 © 2001 Sage Publications, Inc.

93

94

ENVIRONMENT AND BEHAVIOR / January 2001

associated with preference, an examination of particular aspects of this literature raises two related issues. The first concerns the basis for the large variation in preference for different scene types. The second relates to the frame of reference people use when making a simple preference judgment. The aim of the experiment to be reported was to examine these questions and particularly whether the preference differences between scene types are the result of participants using an implicit frame of reference when making the judgment and that this reference frame corresponds to the implicit restorative value of the scene.

FRAMES OF REFERENCE AND LANDSCAPE PREFERENCE

One characteristic of the early work in the landscape preference area was that two foci can be identified in relation to the selection of the examples of landscapes that were judged. A substantial and continuing focus concerned the management of forest resources, and consequently, sets of examples were drawn from the different forest types of interest (see, e.g., Arthur, Daniel, & Boster, 1977; Shafer & Brush, 1977; Vining, Daniel, & Schroeder, 1984). A similarly substantial and continuing focus used diverse sets of examples of many different types of outdoor scenes (see, e.g., Calvin, Dearinger, & Curtin, 1972; Dearinger, 1979; Kaplan, 1987; Kaplan & Kaplan, 1982; Kaplan, Kaplan, & Wendt, 1972). In the first case, the range of scene types was of necessity restricted because of the purpose of the research. In the second, the range was often large, with little attention being paid to the issue of scene type; rather, the emphasis was on diversity and variation in preference across scene types. Because of the particular emphasis associated with each of these foci, little attention was paid to the issue of the types of scenes and particularly to the nature of the psychological categories of scenes derived from regularities in the physical attributes of the environment. An exception to this is the work of Kaplan and Kaplan (1982, 1989). In an extensive series of experiments using a diversity of examples of outdoor scenes, they used multivariate analyses of preference judgments to identify clusters of examples found in different parts of the best-fitting multivariate model. The physical attributes associated with these clusters were then analyzed and used to develop an information-processing model of preference. The clusters identified in this way could be considered as scene types; however, they are preference-defined scene types, which may or may not correspond to scene types based on environmental regularities.

Purcell et al. / SCENE TYPES

95

Subsequent to the work of Kaplan and Kaplan, Purcell (1984, 1986, 1987, 1992) applied a model of preference and other types of affective experience developed in cognitive psychology (Gaver & Mandler, 1987; Mandler, 1984; Purcell, 1982) to environmental experience. This model depends on the relationship between the physical attributes of a scene and the mental representation or schema developed through an implicit learning process based on long-term exposure to the regularities present within the environment. In this model, affective experience depends on differences or discrepancies between the particular example and the relevant schema(s), with the type and intensity of the emotion depending on extent of the difference. This model achieved some success in accounting for variations in preference; however, in considering the results of the research, it became apparent that a number of other factors could have contributed to the results. In common with much of the other research using diverse sets of examples, the participants were asked to judge each one as an example of a specific scene type: a landscape. Effectively, this instruction would activate schemas associated with this particular scene type. and all other scene types would, then, to a greater or lesser extent, be discrepant with this schema. These between scene-type discrepancies, in addition to any discrepancies due to differences between the landscape schema and examples of landscapes, could account for the results. This analysis poses the question of whether the results would change if people were asked to make a preference judgment with the reference frame or context for the judgment being related to the type of scene to which the example belonged. Further analysis of this issue identified another facet to the context or reference frame issue. Scenes are not simply experienced in relation to the type of scene depicted. Each scene also can be seen in terms of its function: what the individual could do there or the activities that would normally be associated with the scene type. To address these issues (Peron, Purcell, Staats, Falchero, & Lamb, 1998; Purcell, Lamb, Mainardi Peron, & Falchero, 1994), we set out to define a set of typical scene types common to two geographic locations: eastern Australia and northern Italy. Initially, we did this using land-use criteria, but subsequently, we assessed the validity of the scene types using groups of people similar to those who would participate in the experiments. In addition, we generated a set of names that could be used to refer to the scene types and then checked their validity in the same way. Groups of participants from the two countries then made preference judgments of the sets of examples from both countries. However, participants were asked to select the name of the scene type they would use to refer to the particular example and to make their judgment using that scene type as the reference frame. In addition, we varied the

96

ENVIRONMENT AND BEHAVIOR / January 2001

reference frame relating to the function of the scene, with different groups being asked to simply indicate their preference for the example or their preference as a place to live and work or to visit on a vacation. By asking our participants to make the judgment by using a descriptor of a scene type they would apply to the example, we would expect that any effect of judging a scene as an example of an inappropriate scene type would be removed. Our use of examples of the same scene types from two different locations, with groups of participants from both locations making the preference judgment for both sets of scenes, allowed us to clarify the source of difference or discrepancy. Because of differences between the two geographic locations in terms of the physical attributes associated with each scene type, we expected that scenes from outside of the geographic location of the participants would be discrepant with existing schemas and would consequently be associated with preference. By having the preference judgment made as a global preference judgment or in relation to a particular function, we were able to assess the contribution of this form of reference frame to the judgment. The experimental results were quite clear-cut. Where participants made a global preference judgment or judged the scene as a preferred place to live and work or to visit on a vacation, there were differences within a scene type depending on the frame of reference. However, the most striking feature of the results was the variation between the different scene types, independent of the frame of reference. That is, the pattern of variation in preference between scene types was the same for the different frames of reference, and this effect was much larger than the effect of frame of reference within a scene type (for details of these results, see Purcell et al., 1994). This variation between scene types was also clearly not due to having to make the judgment on the basis of one scene type (that is, a landscape), because each scene type was judged in terms of a descriptor chosen by the participants as appropriate for the example being judged. Where the comparison was between judgments made of examples of the scene types from within or outside the home environment, a manipulation of the difference of the example to the existing schema within each scene type, differences were found that were consistent with a discrepancy model of preference. However, this did not apply to all of the scene types, and the largest effect in the data was, again, the variation between the different types of scenes, with this variation being similar whether the scenes being judged came from within or outside the home environment (for details of these results, see Peron et al., 1998). Although both the activity frame of reference and the discrepancy between examples can account for differences between examples within the scene types, they cannot account for the large variation in preference between scene types.

Purcell et al. / SCENE TYPES

97

RESTORATIVENESS AS AN IMPLICIT FRAME OF REFERENCE

These results, therefore, pose the question of the basis for the pattern of preference differences between scene types. It has been argued that a global preference judgment is global in the sense of being undifferentiated (Kaplan & Kaplan, 1982). However, it could be argued that such a judgment cannot be made in the absence of any frame of reference and that participants supply one, even if it is not part of their conscious experience. One possible frame of reference could be the restorative value of the scene. This was suggested by the results from the comparison between an overall preference judgment and preferences for the scene types for various activities. The judgment of a scene as a place to visit on vacation contains aspects that are clearly related to the types of items used to assess the perceived restorativeness of environments (Hartig, Mang, & Evans, 1991; Kaplan, 1995; Korpela & Hartig, 1996), and it was this judgment that was most similar to the overall preference judgment. The aim of the experiment to be reported, therefore, was to examine this possibility by comparing overall preference judgments for different scene types with judgments of the same scene types in terms of their restorative value.

METHOD STIMULI

In our previous work (Peron et al., 1998; Purcell et al., 1994), 12 scene types were used. However, in this experiment, we used a subset of 5 scene types drawn from the overall set. This decision was made because of the number of judgments that have to be made to assess the restorative value, preference, and familiarity of a scene. Five scene types from the Italian set used in previous experiments were chosen, with each scene type being represented by two examples, giving a total of 10 stimuli. The 5 scene types were industrial zone, houses, city streets, hills, and lakes; they were chosen to span the full preference range found in our previous work. Five groups of two example scenes were formed, with the two examples in a group being drawn from different scene types. PARTICIPANTS

A total of 100 students (50 males and 50 females) from the University of Padua participated in the experiment. One third of the participants had lived

98

ENVIRONMENT AND BEHAVIOR / January 2001

in Padua from birth, the others for a minimum of 5 years, with the mean residence time being 11 years. Ages ranged between 18 and 29 years with a mean of 22 years. We used five separate groups of 20 subjects (10 males and 10 females) to make judgments of two example scenes from one of the five groupings of the 10 stimuli. JUDGMENTS

Each scene was judged on the 29 items making up the Perceived Restorativeness Scale (the PRS; Hartig, personal communication, July 1997) (see Table 1), together with a familiarity judgment and two measures of preference: the extent of liking for the place and preference relative to all other places where the individual had been. This version of the PRS is presented in Table 1 and includes items relating to an additional subscale, Scope, and to the four subscales used in previous work with the PRS: Being Away, Fascination, Coherence, and Compatibility. However, two items on the scale relate to two of the subscales. Item 7 (There are few hard boundaries here to limit my possibilities of moving about) relates to the Scope and Compatibility subscales, and Item 10 (This place is large enough to allow exploration in many directions) relates to the Scope and Fascination subscales. The familiarity judgment was included to determine the role of familiarity in the preference for and restorative value of different types of scenes, given that it has been argued that familiarity plays a significant role in preference judgments (Kaplan & Kaplan, 1982, 1989). The judgments were made on a 0 to 10 scale with end points of 0 = not at all and 10 = completely.

RESULTS

For each subject, means on each of the subscales for each example were calculated, with Items 7 and 10 being included in the calculations for each of the relevant subscales. Similarly, the mean was calculated for the two preference items. A reliability analysis was performed for each of the subscales using Cronbach’s alpha. The results were .84 for Being Away, .90 for Fascination, .60 for Coherence, .63 for Scope, and .82 for Compatibility. It is apparent that the reliability of the Coherence and Scope subscales is lower than the reliability of the other subscales; this result for Coherence has been found in previous research using the PRS (see Korpela & Hartig, 1996). Although the psychometric properties of the PRS were not the focus of this experiment, we performed a factor analysis to determine whether an

Purcell et al. / SCENE TYPES

99

TABLE 1 Items Making up the Restorative Scale 1. This place is a refuge from unwanted distractions (Being Away) 2. There is a clear order in the physical arrangement of this place (Coherence) 3. This place does not place demands on me to act in a way I would not choose (Compatibility) 4. This place is fascinating (Fascination) 5. Spending time here gives me a break from my day to day routine (Being Away) 6. Following what is going on here really holds my interest (Fascination) 7. There are few hard boundaries here to limit my possibilities for moving about (Scope) 8. The things and activities I see here seem to fit together quite naturally (Coherence) 9. This is a place to get away from the things that usually demand my attention (Being Away) 10. This place is large enough to allow exploration in many directions (Fascination) 11. There is little here to prevent me from doing what I would choose to do (Compatibility) 12. Being here helps me to stop thinking about the things that I must get done (Being Away) 13. This place awakens my curiosity (Fascination) 14. I experience few demands for concentration when I am here (Being Away) 15. Being here fits with my personal inclinations (Compatibility) 16. When I am here I don’t have to focus on things that I’m not really interested in (Being Away) 17. It seems like this place goes on forever (Scope) 18. It is easy to do what I want here (Compatibility) 19. I can find my way around here without trouble (Compatibility) 20. There is much to explore and discover here (Fascination) 21. This place is familiar to me (Familiarity) 22. My attention is drawn to many interesting things here (Fascination) 23. It is easy to see here how things are organized (Coherence) 24. This place has the quality of being a whole world to itself (Scope) 25. The activities that it is possible for me to do here are activities I enjoy (Compatibility) 26. It is hard to be bored here (Fascination) 27. Everything here seems to have a proper place (Coherence) 28. I like this place (Preference) 29. I prefer this place over all other places I have ever been (Preference)

overall PRS scale score based on the mean of the subscales could be used to represent the restorative value of a scene and to examine the relationship of the items on the scale that related to more than one subscale. A two-factor solution was obtained; one factor accounted for 40% of the variance, and a second factor accounted for 2.1% of the variance. Three of the four items making up the Coherence subscale did not load on the main factor, and it was

100

ENVIRONMENT AND BEHAVIOR / January 2001

these items that loaded on the second factor. For those two items that had been found in previous work to load on two subscales, Item 7 did not contribute to the variance associated with either Scope or Compatibility. Item 10, however, was found to explain more of the variance in Fascination than Scope. These analyses, as a result, indicate that the Coherence subscale may not be related to the main factor characterizing the PRS. It is clear that the structure of the PRS needs further examination. However, in terms of examining the relationship between restorative value, preference, and familiarity, it is apparent that the major psychometric characteristic of the scale is that there is one dominant factor made up of four of the subscales, with one small factor based on Coherence. Consequently, the main analyses were performed on overall scale scores based on the average of the five subscales or the major four subscales, dropping Coherence. The results of these analyses demonstrated that including or excluding Coherence did not alter the results of the analysis. The reason for this effect would appear to lie in the limited range in the Coherence scores for the five scene types (range 5.3 to 6.4, with three of the four scene types being in the range 5.7 to 6.4). This compares to substantial variation in the other four subscales: Being Away: 3.6 to 6.4, Fascination: 2.7 to 6.4, Compatibility: 3.1 to 5.7 and Scope: 4.0 to 6.3. Consequently, the main analysis reported is based on the overall scale score calculated as the average of the five subscale scores for each individual for each stimulus. An analysis of variance was carried out to examine the relationship between preference, familiarity, and the PRS and scene type. The analysis was a repeated measures analysis with scene type as a between-subjects effect and the overall restorative scale score, preference, and familiarity as the within-subjects, repeated measures effect. The analysis also included a between-subjects variable related to the sex of the participants to assess whether differences were associated with this factor. The main effect of scene type was significant (F = 29.0, df 4, 190; p = .01), as was the repeated measures, within-subjects effect (F = 160.6, df 2, 380; p = .01) and the interaction effect between scene type and the repeated measures effect (F = 10.8, df 8, 380; p =.01). The effect associated with the sex of the participants and its interaction with the other effects was not significant. Table 2 presents the means and confidence intervals for each of the scene types for the overall restorative scale score, preference, and familiarity. The basic hypothesis being examined in this experiment concerns the relationship between scene type and the three experiential measures of the restorative value of the environment and the preference for and familiarity of the scene type. As a result, the significant main effect found for scene type is not relevant to the basic hypothesis, as it represents differences between scene types averaged across the repeated measures, that is, the three different scales.

Purcell et al. / SCENE TYPES

101

TABLE 2 Means and 95% Confidence Intervals (in parentheses) for the Restorative Scale, Preference, and Familiarity for Each Scene Type

Scene Type Industrial zone Houses City streets Hills Lakes

Restorative Scale 3.6 3.9 4.5 5.9 6.2

(1.2) (1.2) (1.5) (1.6) (1.5)

Preference 1.2 1.9 3.1 4.6 5.8

(1.9) (1.7) (1.8) (2.5) (2.2)

Familiarity 4.6 5.7 8.5 7.5 6.5

(3.5) (2.7) (1.7) (2.7) (3.0)

Preference (Peron et al., 1998) 1.4 2.7 3.6 4.5 5.8

The significant main effect for the repeated measures effect demonstrates that there are differences between the restorative scale, preference, and familiarity. However, again, this main effect is also not directly relevant to this experiment because it represents the relationships between the scales averaged across the scene types. The significant interaction effect, however, assesses the relationship between the scene type and the three scales, which is the central concern of the experiment. The nature of the significant interaction can be assessed by examining Table 2. The means for each scene type in the table were arranged on the basis of increases in the restorative scale score. It is apparent from the table that the order of the preference means is the same as the order of the restorative scale means and that both are quite different from the order of the means for the familiarity judgment. Both the restorative scale and preference means increase, whereas the familiarity means show a quite different curvilinear pattern. As an additional way of illustrating the relationship between the restorative scale, preference, and familiarity, Pearson correlation coefficients were calculated between each of the scales across the five types of scenes, using the data from each individual. Preference and the restorative scale correlate .81, familiarity and the restorative scale correlate .31, and preference and familiarity correlate .32. All of these correlations are statistically significant (p ≤ .01); however, it is apparent that the correlation between preference and restorative scale is much larger than the correlation between familiarity and these two scales.

DISCUSSION

If the restorative value of a scene were the implicit frame of reference within which people make a simple preference judgment, then, as a minimum, it would be expected that preference and the restorative scale should covary, and this

102

ENVIRONMENT AND BEHAVIOR / January 2001

has been demonstrated in the analysis of the results. The means for the different types of scenes in Table 2 indicate that this is the case. The restorative value of a scene type systematically increases, moving from the scenes depicting the industrial zone to scenes depicting lakes. This is mirrored by the change in preference as a function of scene type, and the high correlation of .81 found between the restorative scale and preference confirms this relationship. It is also apparent from a comparison between the means for familiarity and the restorative scale and preference that familiarity is not consistently related to either the restorative scale or preference, and this is reflected in the much lower correlation between this scale and the other two scales (familiarity and the restorative scale r =. 31, familiarity and preference r = .32). The scenes that are most familiar occur in the middle of the restorative scale, and preference ranges, with familiarity decreasing at the highest levels of the other two scales. Familiarity does, however, covary for the lower levels of preference and the restorative scale, which would account for the low but significant positive correlation of .32. This result also appears to indicate that familiarity does not play a major role in preference, contrary to some of the findings of earlier research on preference. Similarly, it would appear that the restorative value of an environment is also not related to familiarity or that the relationship is confined to particular types of scenes. As Kaplan and Kaplan (1989) noted, familiarity may play a complex role in preference, and this would appear to apply to the restorative value of environments as well. It is possible, however, that this relationship between the restorative value of an environment and preference could be the result of the repeated measures design. Participants assessed the two examples of a scene type on the restorative scale items and preference and familiarity. There has been some discussion in the literature that making a preference judgment could effect the restorative scale judgments and vice versa. In the version of the restorative scale used in this experiment, the restorative scale items occur first, followed by the preference judgment and then the familiarity judgment. Each group of participants judged an example of two randomly selected scene types. Consequently, it could be that the judgments of the restorative scale items for the first example then affected the preference judgment for that example and possibly established a restorative frame of reference that then affected the preference judgment for the next example of a different scene type. This effect could occur for all groups of participants, even though the two scene types and the associated examples were randomly selected for each group of participants, because it results from the organization of the items in the restorative scale. Although this argument may have some force, a comparison of the preference means obtained in this experiment with the preference means for the same examples and scene types obtained in our previous

Purcell et al. / SCENE TYPES

103

work, where these preference judgments were made first (Peron et al., 1998, shown in the last column of Table 2), demonstrates that it does not apply in this case. Both sets of preference means vary in the same way for the scene types. Given that the Peron et al. judgments were made in the absence of the restorative scale items, the correspondence between the two sets of means is, in fact, additional evidence for the hypothesis that simple preference judgments, such as those made in the Peron et al. (1998) experiment, may involve an implicit restorative frame of reference. It is also worth noting that the results of the analyses confirm once again the importance of content variables in preference and extend these findings to the restorative value of scenes. The scene types in Table 2 are arranged in order of preference and restorative value. It is apparent that highly preferred and restorative scenes (hills and lakes) are natural, with significant variation in topography, and in the case of the highest level of both, preference and restorative value are associated with the presence of water as well as the other two variables. Conversely, the scene type lowest in preference and restorative value is the industrial scenes, which are predominantly built with little topographic variation and no water present in the scene. Our results, however, also demonstrate that the highest preference and restorative value are associated with natural scenes that show little presence of human-induced change. That is, they could be considered wilderness rather than everyday nature, which has been considered in previous work as fitting best to the theoretical basis for the restorative effects of natural environments (see, e.g., the discussion in Korpela & Hartig, 1996).

CONCLUSIONS

The high positive correlation between the restorative scale and preference found with five different types of outdoor scenes gives support to the hypothesis that the well-documented preference differences between different types of scenes could be the result of participants using the restorative value of the scene as an implicit frame of reference for making the preference judgment. What is clearly needed now is data that assess this relationship for a diversity of other types of scenes in other geographic locations and cultural contexts. Similarly, the role of familiarity in both the preference for and the restorative value of scenes will only become apparent when a greater range of scene types and locations have been examined. However, even if the relationship between preference and restorative value proves to be robust, the finding of covariation between restorativenss and preference does not necessarily

104

ENVIRONMENT AND BEHAVIOR / January 2001

indicate a causal relationship. Other variables could mediate the relationship, and it is, therefore, essential that future research be directed at this possibility. One possibility is contained in a question that these results pose: What characterizes a restorative environment? In other words, what is it in terms of the physical attributes of the different scene types that makes some scenes more restorative than others do? The different scene types clearly vary in their content, with the naturalness of the scene varying between scene types. Similarly, physical variables associated with the complexity and coherence of the scene and with the three-dimensionality of the scene, all variables that have received some attention in the preference literature (Kaplan & Kaplan, 1982, 1989), could be associated with the restorative value of the different types of scenes. Measuring these physical variables associated with the scenes and then modeling their relationship to restorativeness and preference would provide insights into this question of mediating variables in the preference and restorativeness relationship. Clearly, this is an area of considerable significance for future research. We would also suggest, however, that a possible explanation lies in a particular type of geometry—the geometry of fractals—that is associated with natural scenes; this variable may underlie the physical variables discussed above and may be the mediating variable in the restorative and preference relationship. The development of fractal geometry was strongly linked to issues relating to the mathematical description of forms and shapes that are found in nature, such as mountain ranges and coastlines (Mandelbrot, 1983). A considerable amount of subsequent work has demonstrated that a wide range of natural phenomena are fractal (see, e.g., Barnsley et al., 1988). However, the work on fractal geometry has, from the beginning, had an additional dimension. Mandelbrot’s seminal book was titled, The Fractal Geometry of Nature, and it includes an ongoing discussion of how aesthetic experiences associated with natural forms could be related to their fractal characteristics. This theme has been extended to the artificial forms that can be generated using fractal equations, to many different types of natural scenes and representations of them such as photographs, and to traditional art works (see, e.g., Briggs, 1992). However, this discussion linking fractals to aesthetic experience is not based on empirical evidence. Examples of natural scenes and artworks are illustrated, and possible ways in which particular features are based on fractals are discussed. However, changes in the fractal geometry associated with different scene types can be empirically investigated, and we are about to embark on such a project. It may be that variations in both preference and the restorative value of scenes depends on their underlying geometry, with high preference and restorativeness being associated with fractals and low preference and restorativeness being associated with, for example,

Purcell et al. / SCENE TYPES

105

underlying Euclidian geometry typical of built environments. This may be the case, or it may be that variables such a physical complexity or the spatial qualities depicted can be shown to be the physical basis for restorative value and preference. This experiment has demonstrated that at the experiential level, restorative value and preference covary.

REFERENCES Arthur, L. M., Daniel, T. C., & Boster, R. S. (1977). Scenic assessment: An overview. Landscape Planning, 4, 109-129. Barnsley, M. F., Devaney, R. L., Mandelbrot, B. B., Peitgen, H.-O., Saupe, D., & Voss, R. F. (1988). The science of fractal images. New York, Springer-Verlag. Briggs, P. (1992). Fractals: The patterns of chaos. London: Thames and Hudson. Calvin, J. S., Dearinger, J. A., & Curtin, M. E. (1972). An attempt at assessing preferences for natural landscapes. Environment & Behavior, 4(4), 447-470. Dearinger, J. A. (1979). Measuring preferences for natural landscapes. Journal of the Urban Planning and Development Division, 105, 63-80. Gaver, W. W., & Mandler, G. (1987). Play it again, Sam: On liking music. Cognition and Emotion, 1(3), 259-282. Hartig, T., Mang, M., & Evans, G. W. (1991). Restorative effects of natural environment experience. Environment & Behavior, 23, 3-26. Kaplan, S. (1987). Aesthetics, affect, and cognition: Environmental preference from an evolutionary perspective. Environment & Behavior, 19(1), 3-32. Kaplan, S. (1995). The restorative benefits of nature: Toward an integrative framework.Journal of Environmental Psychology, 15, 169-182. Kaplan, S., & Kaplan, R. (1982). Cognition and environment: Functioning in an uncertain world. New York: Praeger. Kaplan, S., & Kaplan, R. (1989). The experience of nature: A psychological perspective. Cambridge, UK: Cambridge University Press. Kaplan, S., Kaplan, R., & Wendt, J. S. (1972). Rated preference and complexity for natural and urban visual material. Perception and Psychophysics, 12(4), 354-356. Korpela, K., & Hartig, T. (1996). Restorative qualities of favorite places. Journal of Environmental Psychology, 16, 221-233. Mandlebrot, B. B. (1983). The fractal geometry of nature. New York: W. H. Freeman. Mandler, G. (1984). Mind and body: Psychology of emotion and stress. New York: Norton. Peron, E., Purcell, A. T., Staats, H. J., Falchero, S., & Lamb, R. J. (1998). Models of preference for outdoor scenes: Some experimental evidence. Environment & Behavior, 330(3), 261-305. Purcell, A. T. (1982) The structure of activation and emotion. Multivariate Behavioral Research, 17, 221-251. Purcell, A. T. (1984). The organisation of the experience of the built environment. Environment and Planning B: Planning and Design, 22, 173-192. Purcell, A. T. (1986) Environmental perception and affect: A schema discrepancy model. Environment & Behavior, 18(1), 3-30.

106

ENVIRONMENT AND BEHAVIOR / January 2001

Purcell, A. T. (1987) Landscape perception, preference, and schema discrepancy. Environment and Planning B: Planning and Design, 14, 67-92. Purcell, A. T. (1992). Abstract and specific physical attributes and the experience of the landscape. Journal of Environmental Management, 34, 159-177. Purcell, A. T., Lamb, R. J., Mainardi Peron, E., & Falchero, S. (1994). Preference or preferences for landscape. Journal of Environmental Psychology, 14, 195-205. Shafer, E. L., & Brush, R. O. (1977). How to measure preferences for photographs of natural landscapes. Landscape Planning, 4, 237-256. Vining, J., Daniel, T. C., & Schroeder, H. W. (1984). Predicting scenic values in forested residential landscapes. Journal of Leisure Research, 16(2), 124-135.