new banner

Services of Scenic Solutions
Completed landscape projects
Papers and Presentations
Beautiful landscapes
Australian and International studies of landscape quality
The science of scenery
Landscape miscellany
Landscape resources
About Dr Andrew Lothian
Contact us
Site map

For other segments of the Science of Scenery, click here:


This section of the web site examines the need for theory and then describes the four principal theories advanced in the area of landscape aesthetics. It also reviews two quasi-theories or models for looking at landscape aesthetics. All four theories are variations of an evolutionary perspective as each assumes that landscape preferences are survival-enhancing. The sections, theories and models are (Click on these):

Need for Theory

Consistent with the development of a new area of intellectual inquiry, the landscape field has been characterised as “rampantly empirical” [Porteous, 1982], lacking a sound theoretical base to guide it. Appleton stated similarly that the “techniques of evaluation are overwhelmingly dominated by empirical methodology, [and] that they could be greatly strengthened if they were underpinned by a more convincing theoretical base” [1975b]. Buhyoff and Wellman [1980] considered that the point has been reached “where [a] theoretically based model development should become a primary goal”.

In a landmark review of over 160 landscape research papers, Zube, Taylor & Sell “identified a conspicuous theoretical void in the majority of the research” [1982]. However, in a subsequent paper, they identified perception research to be based on “a scattering of diverse theoretical origins.” [Sell, Taylor & Zube, 1984]. Using the four paradigms of perception research they had identified, they describe these theoretical origins, which are discussed below. Nevertheless Zube, Taylor & Sell also agree with Appleton that the lack “of a unifying theoretical structure does not allow a rational basis for ‘diagnosis, prescription and prognosis.’”

While the lack of theory is widely recognised, the reason for the void is less apparent. Part of the reason may be that the philosophy of aesthetics and the literature on landscape design and art history have much about aesthetics but notoriously little of a practical orientation which could apply to landscape quality assessment [Dearden & Sadler, 1989]. Also, because of rapid changes to landscapes, some argue that practitioners “were not going to fiddle with theory while the landscape burned.”

Why is it necessary to have a theoretical basis? If the community is concerned about landscape quality, is that not enough? While people’s opinions may be sought about the worth or quality of a landscape, there is no way of making sense of these views without a theoretical construct. Theoretical paradigms can provide managers with the basis for management action, by allowing prediction of consequences following action.

A further reason is analogous to the understanding of the human body that separates a doctor from the person in the street in making a diagnosis; as Appleton aptly put it: “just as the Brisbane wicket after rain used to be said to reduce all batsmen to an equal plane of incompetence, so this absence of aesthetic theory brings the professional down to the same plane as the man in the street.” [1975b] Theory can thus provide a basis for elevating the level of analysis from common to expert. Theoretical perspectives also assist framing problems, in defining what to look for and in what ways to look [Gärling & Golledge, 1989].

Following the comprehensive review that he undertook with Taylor and Sell, Zube stated that the lack of theory and narrow approaches restrict the future growth of the field [Zube, 1984]. The lack of an adequate theoretical base constrains the identification, assessment and protection of landscapes. The “task of theory in landscape aesthetics”, according to Bourassa [1991] “is one of identifying aesthetic laws, if they exist, and of identifying the general characteristics or types of aesthetic rules and strategies.”

Go to top

Habitat Theory - Gordon Orians

Habitat theory is an overarching paradigm within which fit information-processing theory and Appleton’s prospect-refuge theory. Bourassa [1991] suggested the overlap between these two theories. Appleton viewed his theory as having its roots in habitat theory. Habitat theory may be defined as:

“the theory that aesthetic satisfaction experienced in the contemplation of the landscape stems from the spontaneous perception of landscape features which, in their shapes, colours, spatial arrangements and other visible attributes, act as sign-stimuli indicative of environmental conditions favourable for survival, whether they are really favourable or not.” [Appleton, 1975]

In the early 1970s when Appleton wrote his book, information processing theory (insofar as its application to landscape) was in its infancy. G.H. Orians, an evolutionary biologist, and the principal advocate of the theory, states that its biological underpinnings are that:

“natural selection should have favoured individuals who were motivated to explore and settle in environments likely to afford the necessities of life but to avoid environments with poorer resources or posing higher risks.” [Orians & Heerwagen, 1992]

Habitat theory postulates that, because the habitats in which humans are believed to have evolved were dominated by grasslands and scattered trees with water in close proximity, this became a preferred visual landscape for humans. It has been believed that the East African savanna was the cradle of humanity [Leakey, 1963, 1976]. Balling and Falk state that much of our “biological apparatus, most obviously bipedalism, is that of a savanna primate.” [1982].

Research by Rabinowitz & Coughlin [1970] found that there was a general preference for landscapes that were “parklike” or “obviously man-influenced.”

“Mowed grass and scattered large shade trees seem to be the determining factors. Judges may say, ‘This is nice because it looks natural, away from civilization.’ However, the scenes to which they are referring are not in a wild or natural state but clearly ‘landscaped’.”

These environments, the authors suggest, provide feelings of openness and seclusion, or in Appleton’s terms, prospect and refuge. Habitat theory may provide a plausible explanation for the importance of the pastoral landscape, from the Arcadia of antiquity through the paintings of Claude and Poussin and the landscape gardens of Capability Brown to the municipal parks of today. The preference for parklike landscapes is the only landscape form that appears to have endured across the millennia. Balling and Falk ask:

“Are many of the parks and backyards people have so assiduously created wherever they have lived in part an expression of an innate predisposition for the savanna?” [Balling & Falk, 1982]

 Ulrich found in a survey of Swedes and Americans a preference for park-like scenes. These were:

“distinguished by the presence of scattered trees or small groupings of trees, and all had even or fine ground textures. In some cases the scenes had been landscaped and the textures consisted of mowed grass. The even ground textures contained relatively little complexity; rather, the bulk of the complexity consisted of vertical elements - trees and bushes - which stood out clearly against the unambiguous depth “sheet” of the ground surface. “ [1977]

According to Orians the:

“savannas of tropical Africa have high resource-providing potential for a large, terrestrial, omnivorous primate ... In savannas ... trees are scattered and much of the productivity is found within two metres of the ground where it is directly accessible to people and grazing and browsing animals. Biomass and production of meat is much higher in savannas than in forests.” [1986]

Based on this, Orians suggests that:

“savanna-type environments with scattered trees and copses in a matrix of grassland should be highly preferred environments for people and should evoke strong positive emotions.”


“tree shapes characteristic of environments providing the highest quality resources for evolving humans should be more pleasing than shapes characterising poor habitats.” [Heerwagen & Orians, 1993]

acacia tree
African acacia trees

G.H. & E.N. Orians photographed African savanna trees, in particular the Acacia tortulis, and selected trees varying in height/width ratio, height of branches, extent of canopy layers. Photographs were selected to test four hypotheses:

  • Trees with lower trunks should be more attractive than trees with high trunks
  • Trees with moderate canopy density should be more attractive than trees with low or high canopy density
  • Trees with a high degree of canopy layering should be more attractive than trees with low or moderate degrees of layering
  • The broader the tree canopy relative to its height, the more attractive the tree should be [Heerwagen & Orians, 1993]

Measures were taken of each tree canopy’s width and height, tree height and trunk height.  These were converted into ratios of canopy width/height, canopy width/ tree height, and trunk height/tree height. Respondents rated attractiveness of photographs [b & w] of the trees on a 6 point scale. The study found that trunk height, canopy layering and canopy width/tree height ratio significantly influenced attractiveness scores, but the canopy width/canopy height did not have a significant effect.

The most attractive trees [Table1] had highly or moderately layered canopies, lower trunks, and higher canopy width/tree height ratio. Factors such as broken branches, deformed trunks, and highly asymmetrical canopies and indicators of resource depletion depressed attractiveness scores.

Table 1  Comparison of Most & Least Attractive Trees

7 most attractive 7 least attractive     t     p
Mean attractiveness score
Trunk height/tree height ratio
Canopy width/tree height ratio
Canopy width/ canopy height ratio

Source: Heerwagen & Orians, 1993

Interpreting their results, the authors noted that “a low trunk is easier to climb than a high one; a broad umbrella-like canopy affords greater refuge from sun or rain than a narrow, high canopy.” [Heerwagen & Orians, 1993]. The results were considered to support the functional-evolutionary perspective.

Orians and Heerwagen also compared the forms of African savanna trees with maple and oak trees found in Japanese parks and gardens (The choice of Japanese parks and gardens rather than European or North American was not explained). Comparing three morphological differences - height vs canopy width, trunk height vs total height, and canopy depth vs canopy width - they found close similarities:

“Garden conifers are highly modified by pruning them to grow broader than tall; trunks are trained to branch close to the ground; foliage is trimmed to produce a distinct layering similar to that of a number of savanna species.”.

While suggesting that achieving a growth form similar to that of savanna trees was a criterion subconsciously employed by Japanese gardeners, Orians recognised that many other factors also have had an influence [Orians, 1986].

In another study, Heerwagen & Orians sought evidence for Appleton’s prospect and refuge among landscape paintings and the Red Books of Humphrey Repton, the 18th century English landscape architect. Their analysis of Repton’s Red Books also examined whether he created savanna-like scenes. These books illustrated the “before” and “after” appearance of properties, showing the effects of his landscaping. Examination of 18 designs found that Repton frequently moved trees out into open space, thereby creating an uneven wood edge, a feature characteristic of savanna environments. In his book, The Art of Landscape Gardening, Repton noted that too many trees “make a place appear gloomy and damp.”.

According to Sommer and Summit, research on tree preferences in Argentina, Australia and United States found that:

“respondents preferred canopies to be moderately dense and trunks that bifurcated near the ground. Trees with high trunks and skimpy or very dense canopies were considered to be least attractive by all these groups, findings considered to be consistent with the savannah hypothesis” [Sommer & Summit, 1995].

Sommer and Summit used computer drawn images of tree shapes to test preferences with variations in height and width. They found preferences for large canopies [chi sq = 195.7, p < 0.001], low trunk height and thin trunk thickness [both p < 0.001], the first two properties being consistent with savanna hypothesis and the third [trunk thickness] being irrelevant.

Both Balling and Falk [1982] and Lyons [1983] assessed the preferences for a range of environments illustrating savanna, deciduous forest, coniferous forest, tropical rain forest and desert. Both found savanna to be the most preferred of the five biomes. They found that preference for savanna was highest among the age 8 - 11 year olds after which it slipped behind deciduous and rain forest and, in Lyons’ study, behind rain forest. Balling & Falk found that overall preference for natural environments changed as a function of age [Ibid, 16], [F = 89.62, df 5, 492, p <0.001].

Figure1 indicates the shift in preferences for savanna with age. While the scores differ between the studies, the pattern is similar: high scores among the young that fall progressively with age, stabilising in adulthood. 


Source: Balling & Falk, 1982; Lyons, 1983. Note: Lyons study results significant at p < 0.05; Balling & Falk at < 0.001
Figure 1 Comparison of Preferences for Savanna by Age

Both found the preference for savanna was strongest when a lush green savanna was used in preference to a drier African-like savanna. The difference was so striking that Lyons dropped the lush green savanna. The use of the greener savanna in the Balling and Falk study probably accounts for the higher ratings.

 While Balling and Falk believed the results provide “limited support for the hypothesis that people have some innate preference for savanna-like environments”, Lyons disputed this on the basis that the preference for savanna could be related to its familiarity for children who play in savanna-like parks and backyards. Commenting on the functionalist-evolutionary perspective she noted:

“This perspective is also plagued by the same dependence on optimality theory that is evident in much of biological evolutionary theory; it does not recognize that natural selection is not precise, that the current function of a structure cannot be used to infer its adaptive origin, and that some structures or processes that affect landscape preference may in fact be maladaptive but persists because of the correlational structure of the human genome."

Woodcock [1982] also examined preferences for three biomes: rain forest, savanna and mixed hardwoods and found the hardwood to be the most preferred [rainforest 2.83, savanna, 3.06, dense hardwood with underbrush, 3.04, open hardwood with open ground, 3.73]. It is also possible that this may be due to familiarity as suggested by the Kaplans [1989].

Fenton [1985] analysed the underlying dimensions of meaning or content that individuals use in discriminating natural settings. He found that the majority of participants preferred scenes characterised by: open grasslands, verdant, water, natural, and with pathways. He viewed these findings as supporting the Kaplans’ theory, but they also lend support to Orians’ habitat theory.

Schroeder [1991], studying preferences for scenes in an arboretum in Chicago, found natural deciduous wood scenes, large trees, and water attracted the highest ratings but scenes of trees and lawn - the classic pastoral landscape, were less preferred.


Source: Schroeder, 1991
Figure 2 Preferences of Groups for Arboretum Scenes

Among the evidence cited to support habitat theory is the observation that no archaeological evidence has been found to indicate early human occupation of dense forest, rainforests or deserts [Isaac, 1980]. Use of fire by indigenous people, including the Australian Aborigines and the North American Indians, encouraged the development of savanna-like vegetation. While the purpose of this was to create favourable conditions for game, it raises the question whether it was unconsciously directed to create a preferred savanna-like landscape. In both cases the cessation of fires after European settlement resulted in the savanna appearance gradually being lost.

Denevan [1992] suggested that in 1492 the native American landscape was a humanised landscape and that with the decimation of the Indian population by disease and war, the vegetation was re-established. “A good argument can be made that the human presence was less visible in 1750 than it was in 1492.” Similarly Bourassa quotes the archaeologist, Dr Rhess Jones, that after the Aborigines “had either died or had been removed ... soon afterwards it was noticed that the plains were becoming filled with sour grass and light scrub so that it was becoming difficult to graze sheep on them, the attempt being abandoned with great financial loss about 10 or so years later.” [Bourassa, 1991] The loss of pastoral landscapes was also apparent in many other locations.

Orians [1980] cites the perceptions of early explorers in North America who seemed to prefer savanna-like landscapes, although this may be to provide grazing land and reduce hiding opportunities for natives. Bourassa notes that similar preferences were apparent among explorers and settlers in Australia and New Zealand. In his book, Future Eaters [1994], Tim Flannery included a chapter titled “Like Plantations in a Gentleman’s Park”, in which he wrote of the settlers’ efforts to transform the Australian landscape into an English landscape.

Many early paintings of the Australian landscape also displayed park-like environments. Favoured scenes among painters were pastoral landscapes, environments which also made for good grazing land and which did not require clearing to be productive. By contrast, Bernard Smith refers to von Guerard’s paintings of virgin forest that “amply convey the depressing effect so frequently mentioned by travellers and settlers.” [1971]. Anthony Trollope, the English novelist, toured Australasia in the 1870s and wrote, “the fault of the Australian scenery is its monotony.” [1873].

Tim Bonyhady identifies a triad of images portrayed by the 19th century artists: an “antipodean arcadia untouched by European settlement and occupied only by Aborigines enjoying a bountiful existence” [1985], a pastoral arcadia occupied by squatters and their sheep, and a magnificent wilderness, as yet untamed.

While there are findings and anecdotal evidence supportive of the habitat hypothesis, these are not definitive. Lyons’ alternative explanations of familiarity may account for the preferences found by Balling and Falk’s study.

Go to top

Prospect and Refuge Theory - Jay Appleton

Jay Appleton’s prospect-refuge theory has become one of the most widely quoted landscape theories. It derives its inspiration from both habitat theory and information processing theory. Hudson described it as a “seminal contribution” [1992]. Appleton, a geographer at the University of Hull, England, described the theory in The Experience of Landscape [1975]. The book’s name derives from the view of the philosopher, John Dewey, that beauty lay neither in beautiful objects nor in the eyes of the beholder but rather in the relationship between the individual and the environment - what Dewey called ‘experience’. Such experience covers both the habitat theory and information processing theory that aesthetic satisfaction from landscapes derives from their favourability for survival [1975a]. The title of Dewey’s book, Art as Experience (1934) may have inspired Appleton’s title.

In King Solomon’s Ring [1952], Konrad Lorenz wrote of seeing without being seen, which relates to habitat theory. Appleton built on this, arguing that a landscape need only provide the appearance of satisfying survival needs. Certain sign-stimuli provided by the landscape comprise the core of Appleton’s prospect-refuge theory. He termed the sign-stimuli that provide opportunities to see a prospect while those which provide an opportunity to hide he termed refuge. Appleton summarised his theory thus:

“Habitat theory postulates that aesthetic pleasure in landscape derives from the observer experiencing an environment favourable to the satisfaction of his biological needs. Prospect-refuge theory postulates that, because the ability to see without being seen is an intermediate step in the satisfaction of many of those needs, the capacity of an environment to ensure the achievement of this becomes a more immediate source of aesthetic satisfaction.”

Appleton developed the imagery and symbolism of the theory. Prospects can be direct or indirect and include panoramas and vistas while refuges can be classified by function [e.g. hides and shelters], by origin [natural or artificial], by substance [in the earth such as caves or in vegetation], by accessibility and by efficiency. One senses that some of these are classification for classification’s sake but Appleton is nothing if not exhaustive in the development of his theme.

He examined and classified hazards, surfaces and related components, discussed landscapes which are dominated by prospect, refuge or hazard [pp 146 - 168], the place of man in nature [pp 169 - 191] and then reviewed prospect and refuge in  parks and gardens, in architecture and urban design, painting, film, literature [pp 192 - 219] and the application of prospect-refuge theory to the landscape gardens of Capability Brown, Repton and le Nôtre’s Versailles [pp 220-8]. He commented on fashion and taste [pp 220 - 237] and finally described the application of the theory to case studies of landscapes in several countries. [pp 238 - 256].

Over a decade later, Appleton (1988) described how he developed his theory:

“I was looking for a simple model that could relate the idea of preference to a typology of landscapes through the medium of the biological and, more particularly, the behavioural sciences.”

The theory potentially offers an explanation to the perennial question of why people climb mountains. The answer is not “because it’s there” but rather because the mountain represents the best prospect available and, hence, being on top of it enhances survival. The fact that this may lead people to climb very high mountains and to even be killed in the attempt does not negate this hypothesis, it merely suggests that optimality applies in the selection of mountains to provide prospects and that high mountains may actually be sub-optimal for this purpose.

In a Spanish study, Abelló, Bernaldez & Galiano [1986] found preferences for forest landscapes, a preference for fertility and plant vigour, some pattern or rhythm, and a structural legibility in winter defoliation [Ibid, 168]. The survival-promoting preferences tend to support Appleton’s thesis: they “correspond either to signs indicating environmental virtues (fertility and plant vigor healthy biomass) or hazards (environmental hostility present in defoliated wintry vegetation)...”.

Using a very limited sample of four participants [including the authors], Clamp and Powell [1982] sought to test Appleton’s theory by rating 40 panoramas of landscapes for landscape quality, prospect, refuge, hazard, and the balance of prospect and refuge. The authors calculated that, although the quality ratings correlated well, there were no significant correlations between preference and prospect-refuge balance [p < 0.001]. (A sample of 4 is usually regarded as insufficient for correlations) Some correlation was obtained between preference and prospect. They found a significant negative correlation between prospect and refuge - the finding is not surprising as something that provides good prospect is unlikely to be a good refuge. Overall though, the study failed “either to support conclusively or to negate the central claim of [the] theory” and “despite every effort [by the judges they] remained unconvinced that they were tapping some underlying perceptual force.”

Orians suggested that scenes with a high proportion of prospects compared with refuges would be favoured as familiarity of the observer increases and the risks they present decrease accordingly [1986, 9]. He observed that closed forests are deficient in prospect while desert and grassland scenes are deficient in refuge.. Savannas, by contrast provide a good combination of prospect and refuge. Elsewhere, Orians and Heerwagen [1992] suggest that Appleton’s theory means that an environment judged pleasant will be one with a balance between prospect and refuge opportunities, with screening elements to provide privacy and variability in desired levels of intimacy in a space.

Heerwagen & Orians [1993] tested the evidence for prospect and refuge in landscape paintings, by examining gender differences in preferences and by examining the before and after pictures by the English landscaper, Humphrey Repton, and by the painter, John Constable. These are summarised below.

Sunsets in Landscape Paintings

Based on a assumption that paintings of sunsets represent refuge symbolism, it would be expected that artists would include references to places in which people could spend the night. Out of 46 paintings of sunsets and sunrises [including many by Frederick Church], 35 were sunsets and 11 were sunrises indicating they believed that “the information provided by a sunset is much more valuable and requires more urgent attention than ... a sunrise.” [Ibid, 148] The sunset paintings scored very highly in refuge symbolism: 66% scored highly in refuge compared with 9% for sunrises [c2 = 10.89, p = 0.004]. Sunset paintings had more built refuges whereas sunrise paintings had very few. Paintings that included a built refuge also included additional refuge symbols: 46% had a light in the window, 12% had smoke from the chimney, while 7% had both a light and smoke.

Gender differences

Their hypothesis was that females find refuges more attractive: “a greater affinity for enclosure and protected places than do males” due to pregnancy and childcare, as well as protection from the elements, which drain energy. To avoid being trapped or being taken by surprise, an open refuge would be advantageous. Content analysis of 108 landscape paintings, painted by both male and female artists [52 by females, 56 by males] was used. Prospect symbolism included open landscapes, opportunities for views [hills, mountains, rock outcrops], and a view of the horizon at least half the width of the painting. Refuge symbolism included houses and vegetative cover, especially in the foreground. In summary:

  •  Women’s paintings: nearly half were high in refuge symbolism compared with 25% for men’s paintings [c2 = 6.89, p = .03]. 75% had no horizon or peephole, these being symbolic of prospects.
  • Men’s paintings: nearly half were high with prospects compared with 25% for women’s paintings [c2 = 12.07, p = .002]. Nearly 75% had moderate-high prospect symbolism compared with less than half for women’s paintings. The horizon was more than half the width in 58% of paintings compared with 14% of women’s paintings.

Before and After Scenes

Heerwagen & Orians examined the before and after designs of Repton and Constable, the former for his landscaping of properties and the latter of his sketches for later paintings. In 18 scenes Repton enhanced the refuge and prospect character of the properties by adding copses of trees at the water’s edge  which increased refuge and by removing trees to open views to the horizon which increased prospect.

Examination of nine of Constable’s sketches and paintings indicated that he frequently altered the vegetation to open views to the horizon or to make refuge features such as houses more conspicuous. In six of the pairs he enhanced the refuge conditions by adding buildings and changing vegetation.

The findings by Heerwagen & Orians support the prospect and refuge symbolism as an unconscious organising attribute.

Researching forest and field environments, Herzog [1984] used factor analysis to identify three dimensions: unconcealed vantage point, concealed vantage point, and large trees. The parallels with refuge and prospect are obvious. Both the unconcealed and concealed vantage points were moderately well liked with similar ratings of 3.27 and 3.39 on 5-point scale, suggesting little difference in the preferences for each type. He found stronger preferences for large old trees [3.79], which provided a significantly higher rating [p < 0.05]. When these trees were viewed in combination with pathways, ratings of 4.0 were obtained. Herzog speculated that this may be due to the large old trees providing an “especially pleasing effect as pathway border elements”, - an artistic explanation but it might also suggest that the combination of tree and path provide ideal refuge and prospect combinations. Herzog was aware of Appleton’s work, but confined the implications of the study to Kaplan’s theory.

In a study of waterscapes, Herzog [1985] referred to Appleton’s prospect as an affordance in Gibson’s [1979] terms (affordances are the functional use of landscape elements rather than their form, colour and other attributes) , but did not analyses his findings in these terms. He found preferences were, in order [5-point scale], mountain waterscapes [3.99]; large water bodies [3.28]; rivers, lakes and ponds [3.11]; and swampy areas [2.13]. He found swampy areas to be distinguished by low spaciousness [2.45] while large water bodies were distinguished by spaciousness [4.11] and coherence [3.66]. Spaciousness could be equated with prospect, as both denote similar qualities of openness of view. The mountain waterscapes were high in spaciousness and would also be expected to be high in prospect, while swampy areas were low in spaciousness and would also be expected to be low in prospect [but possibly high in refuge, which tends to be ranked negatively in preferences].

Herzog & Smith [1988] examined canyons and urban alleyways to examine Appleton’s concept of hazard and how this related to Kaplan’s predictor variables of mystery. Overall, they found that “both danger and mystery predict preference, the former negatively and the latter positively."

Hull & McCarthy [1988] used scenes of the Australian bush to assess the impact on preferences of wildlife in scenes. Three dimensions were identified: water, enclosure and concealed view, the latter corresponding, they acknowledged, with Appleton’s theory. In a concealed view, foreground vegetation concealed the view but not enough to block views to the middleground or background.

Nasar et al [1983] examined the preferences expressed from two locations in a city park. At each location the observer viewed the scene from a protected position [enclosed] and an unprotected position [Figure 3]. They assessed the scene on a nine bi-polar adjective scale [e.g. repelling-inviting, relaxed-tense].


Source: Nasar et al, 1983
Figure 3  Interactive Effect of Refuge and Gender on Preferences

They found that the open views were regarded as safer than closed views [F = 8.18, df = 1, 56; p < 0.01], which accords with Appleton’s theory. However they also found that females preferred the enclosed observation point to the open one, while the opposite applied to males [F = 3.73, df = 1, 56; p = 0.06]. The notion of males preferring viewing points with less refuge is contrary to Appleton’s theory.

Strumse’s [1996] finding of higher preferences for green, grassy fields among women than men [males 2.99, women 3.22; 5-point scale] could reflect a preference for the “open and well defined settings, which most probably induce feelings of security." Such landscapes offer good prospects in Appleton’s terms.

Woodcock [1982] assessed preferences for three biomes [savanna, rain forest and hardwoods] on the basis of six affordances, including primary and secondary prospect, and primary and secondary refuge. The primary prospect is a photo taken from a high vantage point showing the surrounding landscape while the secondary prospect shows a good vantage point; similarly the primary refuge is of a photo which indicates that it was taken from a concealed location, whereas the secondary refuge only indicates good refuges in the landscape. Woodcock found prospect to be positively related to preference [0.55] while refuge appeared to be negatively related [-0.59], an unexpected result which led him to propose additional predictors including agoraphobia and claustrophobia.

Overall the evidence is not compelling for Appleton’s theory and indicates that some refinement may in order. While prospects generally correlate with preference this may derive from the appeal of mountains. Refuges are generally regarded negatively. A strong dichotomy by gender in preferences for prospect and refuge appears present - males preferring open prospects, females preferring safe vantage points. While Appleton regards the balance between prospect and refuge as important, few studies have attempted to tackle what this balance might be.

Kaplan’s concepts of coherence, complexity, legibility and mystery appear to have some overlap and parallels with Appleton’s prospect and refuge, for example prospect and legibility, refuge and mystery, and this could be explored further. 

Appleton’s theory has been described as a “sociobiological account of aesthetic value” [Carlson, 1992, 79] while Bunkse [1977, 150] described it as “hide and seek aesthetics”. Bunkse considered that the theory “seems to answer many unanswered questions”, including the human preference for natural habitats rather than artificial ones, and in treating the vast differences in French and Japanese gardening styles as attempts to fulfil innate, biologically determined preferences. Appleton considered that cultural differences can be explained by their biological underpinnings, a view not universally shared; e.g. Bourassa: “...arguments such as Appleton’s are rather extreme assertions of a biological basis for aesthetics...” [1991]. Jeans stated [1977]: “The survival of primitivist urges in man, like territoriality, is so overlaid by cultural accretions and modifications that it seems uselessly oversimplistic to seek to apply them to human behaviour.”

Bunkse also questioned the theory’s ability to deal with ambiguities, such as whether darkness is a prospect or refuge, and he cast doubts about its reliance on innate drives saying [1977]:

“It cannot be denied that a good deal of human behaviour can be compared with animals, but as a species we have developed our own unique traits which can be understood only through direct study of humans. Such understanding must be couched not only in terms of biological drives analogous to those in animals, but also in terms of human imagination and the ability to apprehend the self in the environment, and the will to act originally.”

Several reviewers have observed that Appleton’s theory, which suggests that each scene has to be broken down into its prospect and refuge symbolism, “is reductionist in the extreme” [Bunkse, 1977]. Jeans [1977] described it as “ridiculously reductionist” while Tuan described it as “a tour-de-force of reductionism” [1976]. Ulrich considered Appleton’s theory, in which elements are seen to have actual or symbolic survival significance, to be “a rather extreme, ethologically based adaptive position” [Ulrich, 1983].

In 1991, Appleton published The Symbolism of Habitat: An Interpretation of Landscape in the Arts which extended the theme of The Experience of Landscape to the arts. Today, Appleton’s concepts are used consciously by landscape designers [Frey, 1986]. They are cited in site planning text-books and are used in the analysis of literary landscapes and architecture [Hudson, 1992, 56, and Hudson, 1993].

While there is a considerable level of support for Appleton’s theory, it lacks strong supporting evidence. The findings of studies suggest the need for further elaboration and consideration of the theory.

Go to top

Affective theory - Roger Ulrich

Affective theory considers that natural settings and landscapes can produce in their viewers, emotional states of well-being that can be detected through psychological and neurophysiological measures. The main proponent of the theory is Roger Ulrich, originally a geographer at the University of Delaware and more recently with the College of Architecture at Texas A & M University.

Affect is used by Ulrich synonymously with emotion and include feelings such as pleasantness, calm, exhilaration, caution, fear and anxiety [Ruddell  et al, 1989] but excludes drives such as thirst and hunger [Ulrich, 1983]. Although it is measured on a like-dislike dichotomy, it has also been shown to be highly correlated with scales such as beautiful - ugly or scenic quality scales [Ulrich, 1986].

The affective model of preference is based on the premise that emotional [i.e. affective] responses to landscapes occur before cognitive information processing. With the development of cognitive psychology in the 1960s, affects were regarded as products of cognition [i.e. they are post-cognitive]. In a widely quoted paper, Feeling and thinking, preferences need no inferences, Zajonc [1980] argued against the prevailing doctrine that affect is post-cognitive and provides experimental evidence that discriminations [like-dislike] can be made in the complete absence of recognition memory. He concludes that affect and cognition are:

“under the control of separate and partially independent systems that can influence each other in a variety of ways, and that both constitute independent sources of effects in information processing.”

Ulrich also cites evidence in support of affect being precognitive [Ulrich, 1986, Ulrich et al, 1991]. Ruddell, et al consider that the affective state “heavily influence the subsequent cognitive appraisal of a setting as contributing to or detracting from personal well-being.”

Based on this premise, Ulrich constructed a model of affective reactions preceding cognition but both influencing the post-cognitive affective state and actions that then arise [Ulrich, 1983]. He termed the framework a psycho-evolutionary theory, where the positive emotions and physiological effects have survival benefits.

In contrast to the Kaplans’ cognitive perspective, Ulrich proposed that:

“immediate, unconsciously triggered and initiated emotional responses - not ‘controlled’ cognitive responses - play a central role in the initial level of responding to nature, and have major influences on attention, subsequent conscious processing, physiological responding and behavior” [Ulrich et al, 1991].

He also suggested that an  “evolutionary perspective implies that adaptive response to unthreatening natural settings should include quick-onset positive affects and sustained intake and perceptual sensitivity.”

Basic to Ulrich’s framework is that of adaptive response, adaptive meaning the wide array of actions and functioning which can foster well-being [Ulrich, 1983]. Adaptive behaviour may, for example, comprise staying and viewing an attractive scene or setting out to explore it.

Ulrich [1979] tested participants' feelings before and after viewing slides of urban and natural scenes. The results [Figure 4] indicates that urban scenes generally resulted in more negative feelings [e.g. one grew sadder, less elated, less friendly], whereas the opposite occurred after viewing the nature slides.


Source: Ulrich, 1979
Figure 4  Affect Scores Before and After Slides

Negative feelings were lessened and positive feelings became more positive [p < 0.005] from viewing nature scenes. Ulrich showed that the variation attributable to slide content was highly significant [p = 0.002] and concluded that the importance of visual landscapes is not confined to aesthetics, but that they also give rise to emotional states, urban scenes having a negative effect and the nature scenes positive.

In a second study, Ulrich [1981] used psycho-physiological measures to assess the effect of viewing slides of nature with water, nature with vegetation, and urban environments with neither water nor vegetation. He measured alpha waves and heart rates and asked subjects to rate their feelings using semantic ZIPERS scale before and after viewing the slides. Alpha waves reflect brain electrical activity. High alpha amplitudes indicate lower levels of arousal and of wakeful relaxation while anxiety is related to high arousal and low alpha amplitudes. Rapid heart rates reflect strong emotions such as anxiety or fear [Ulrich, 1979]. The ZIPERS scale assesses feelings on five factors: fear arousal, positive affect, anger/aggression, attentiveness, and sadness. A 5 pt scale is used for each.

He found:

  • Attentiveness declined but less so for water scenes [p < 0.001]
  • Sadness increased markedly from viewing urban scenes but only slightly for vegetation and was constant for water - the difference between the influence of urban and water scenes was highly significant [p = 0.005] but less so between urban and vegetation scenes [p = 0.07]
  • Fear arousal emotion increased slightly with urban scenes, decreased slightly with vegetation and declined more sharply with water [urban/water difference p < 0.02]

The physiological measures showed that alpha amplitudes were consistently higher when viewing vegetation than urban scenes with water scenes lying between these [p < 0.05]. The significantly higher results for vegetation were cited as one of the most important findings of the study and support “the conclusion that the subjects felt more wakefully relaxed while viewing the vegetation as opposed to urban scenes”. Heart rates were generally higher while viewing either water or vegetation compared with urban scenes - water 71.3 beats/minute, vegetation 71.1, urban 70.2 [p < 0.20]. Ulrich concluded “people benefit most from visual contact with nature, as opposed to urban environments lacking nature, when they are in states of high arousal and anxiety.”

Ulrich [1984] reported on investigations of the recovery of patients in a hospital, comparing patients whose rooms viewed a blank wall with those who could see trees [Figure 5]. The patients had undergone cholecystectomy [gall bladder] operations. The study found that those who viewed the trees had shorter stays in hospital: 7.96 days vs 8.70 days [T(17) = 35, z = 1.965, p = 0.025], took fewer analgesics sand received fewer negative evaluative comments in nurse’s notes: 3.96 per patient for those facing wall compared with 1.13 for those facing trees [T(21) = 15, z = 3.49, p=0.001].


Source: Ulrich, 1984
Figure 5  Analgesic Doses per Patient - wall & tree views

The analgesic doses did not vary significantly between the two groups for the first day or the last days but for days 2 - 5 the difference was statistically significant [T2 = 13.52, F = 4.30, p < 0.01]. The results imply that “hospital design and siting decisions should take into account the quality of patient window views.” Parsons [1991] considered the results could reflect the differences in complexity between a brick wall and a stand of trees.

Ulrich also found that individuals shown scenes of cities with trees and other vegetation showed significantly reduced  feelings of fear and increased positive feelings of affectation and delight, compared with individuals shown scenes of treeless city scenes [Ulrich, 1979].

Ulrich et al [1991] extended physiological measures to include skin conductance, pulse transit time, muscle tension and heart period. Participants were first tested, they then viewed a ten-minute stressful video [on workplace accidents], and then viewed a second ten-minute video showing everyday outdoor settings - two natural [vegetation and water] and four urban. Pair-wise tests showed that, following viewing natural scenes, positive affect scores increased significantly compared with either the pedestrian mall [p < 0.01] or traffic [p < 0.001]. Results from the four physiological measures showed that the nature scenes reduced stress, indicating their “greater recovery influence.”. The study also found that nature scenes resulted in more rapid recovery from stress, suggesting that even momentary viewings of trees through a window can have benefit.

An early study using eye pupillary dilation as an autonomic measure of aesthetic reaction was undertaken by Wenger and Videbeck [1969]. Applying the technique to both campers and non-campers they found that, although the test provided a reliable pattern of differences between the two groups, the results were opposite of their expectations! On the basis of this finding, the authors concluded that another autonomic measure might be preferable and that the information processing hypothesis may better explain the observed pupillary movement.

Parsons [1991] noted that, although there is no direct empirical evidence supporting Ulrich’s theory, the sensory model of emotions by LeDoux and Henry’s model of endocrine responses in stressful situations “constitute prima facie evidence for the existence of subcortical ‘hardware’ and processing which is supportive.” (LeDoux, J.E., 1986. Sensory systems and emotions: a model of affective processing, Integr Psychiatry, 4, 237 - 248. Henry, J.P., 1980. Present concept of stress theory, in E. Usdin, et al, [Eds], Proceedings of the Second International Symposium on Catecholamines and Stress, Sept 1-16, 1979, Czechoslovakia, Elsevier) He considered that the “immediate affective responses to environments may influence environmental preferences ... and trigger physiological processes that can influence the immune system, and thereby, physical well-being.”

Overall, Ulrich’s research findings provide support for his theory that “immediate, unconsciously triggered and initiated emotional responses - not ‘controlled’ cognitive responses - play a central role in the initial level of responding to nature” [Ulrich et al, 1991].

Go to top

Information Processing Theory - Stephen and Rachel Kaplan

During the 1960s and 1970s environmental psychologists focussed attention on the perception of the environment. Of particular relevance to landscape is the work of Stephen and Rachel Kaplan of the University of Michigan who applied the information processing approach to landscape aesthetics to explain the interactions between humans and the landscape.

The Kaplans hypothesise that "the perceptual process involves extracting information from one's environment." [Kaplan, Kaplan & Brown, 1989] They suggest that humans seek to make sense of the environment and to be involved in it. They identified four predictor variables, two of which (coherence and legibility) help one understand the environment and the other two (complexity and mystery) encourage its exploration [Table 2].

Table 2 Kaplans' Predictor Variables

Making sense
Being involved
The visual array
Making sense now
Orderly, “hangs together”
Repeated elements, regions
Being involved immediately
Richness, intricate
Many different elements
Future, promised
Three-dimensional space
Expectation of making sense in future
Finding one’s way there & back
Expectation of future involvement
Promise of new but related information
  • Coherence is the ease of cognitively organising or comprehending a scene - “good Gestalt”. It involves making sense of the scene. It includes factors which make the scene more comprehensible - to organise it into a manageable number of major objects and/or areas. Research indicates that people hold onto information about scenes in chunks and that up to five can be retained in the working memory. A scene with about five major units will be coherent. Repetition of elements and smooth textures help to identify an area. Changes in texture or brightness should correspond with an important activity in the scene - where it does not, the scene lacks coherence.
  • Complexity is the involvement component - a scene's capacity to keep an individual busy, i.e. occupied without being bored or overstimulated. Often referred to as diversity, variety or richness it used to be regarded as the single most important factor. The Kaplans describes it as how much is “going on” in the scene - a single field of corn stretching to the horizon will not have the same level of complexity as many fields of many crops on undulating land with hedgerows and cottages. The more complex scene will tend to be preferred to the simple.
  • Legibility is the ability to predict and to maintain orientation as one moves more deeply into a scene. It entails “safety in the context of space” [Kaplan, 1979] and is similar, though much broader, to Appleton’s concept of refuge. Legibility, like mystery, involves an opportunity to promise to function, to know one’s way and the way back. It thus “deals with the structuring of space, with its differentiation, with its readability.” Legible scenes are easy to oversee, to form a mental map. Legibility is enhanced by distinctive elements such as landmarks, smooth textures, and the ease of compartmentalising the scene into parts. While coherence focuses on the conditions for perceiving the scene, legibility is concerned with movement within it.
  • Mystery is the promise that more information could be gained by moving deeper into setting, e.g. a trail disappearing, a bend in a road, a brightly lit clearing partially obscured from view by foliage. New information is not present but is inferred from what is in the scene, there is thus a sense of continuity between what is seen and what is anticipated. “A scene high in mystery is one in which one could learn more if one were to proceed further into the scene.” The Kaplans used the term “mystery” reluctantly because they could not find a more suitable term. A better term might be “anticipation”.

In their book, The Experience of Nature [1989], the Kaplans described the studies that contributed to the development of their theory.

An early study, Kaplan et al [1972] focussed on the single factor of complexity and found a 0.37 correlation with preference. A second study [R. Kaplan, 1975] found a correlation of 0.62 between complexity and two new variables, mystery and coherence. However the correlation between complexity and preference, when assessed independently, was -0.47, in contrast with the original +0.37. She put this down to content, the later study being of urban scenes rather than of nature. Using regression analysis, the R2 for the three informational factors was a promising 0.49, indicating that together they accounted for around half the variance. Mystery was particularly significant (r = 0.56), coherence slightly weaker (0.33), and complexity a negative factor (-0.39).

Coherence and complexity are considered to involve minimal analysis, whereas legibility and mystery require more time and thought. Scenes of high preference tend to be those with legibility and mystery; coherence and complexity help create the scene, but high levels of these do not necessarily result in high preference.

Through the 1980s, further studies by the Kaplans, Herzog, Anderson and others reinforced and gave coherence to the definition of the informational variables. Following their review of over a decade’s research, the Kaplans concluded:

  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 in urban scenes).
  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.
  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.

Abello, Bernaldez & Galiano [1986] concluded from their analysis of forested landscape preferences that plant fertility/ vigour factor was a key factor in preference followed by the strong expression of pattern/ rhythm/recurrent texture of landscape elements. Factor analysis indicated correlations of -0.84 and -0.89 of these respectively with the factors they identified. The authors acknowledged that the results lend support to an evolutionary or socio-ecological basis of landscape aesthetics including Kaplan’s “cognitive characteristics related to predicability (pattern recurrent textures) and meaning (legibility of structures, capacity of seeing through barriers).”

Ed Anderson’s [1978] study of forest management assessed informational factors for professional, resident and student groups. Table 3 summarises these factors as predictors of preference for these groups. All of the factors were consistent across all groups with the exception of mystery, which played a negligible role for the preferences of professionals. Coherence and mystery were the best predictors of preference for residents and students.

Table 3   Informational Processing Factors as Predictors of Preference for Groups

Factor Professional Resident Student
- 0.18
- 0.15
- 0.22
- 0.13
- 0.38
- 0.30

Source: Anderson, E., 1978. p<0.05

Brown & Itami [1982] proposed a model that related scenic resource values to landscape preference components as defined by the Kaplan model. The Brown & Itami framework comprises two inter-related systems - the natural (land form) & cultural (land use). These describe the physical components. Landform reflects “immutable“ components and the cultural system is reflected by land use and land cover pattern.

Kaplan model:


Making sense
Visual array
3-D space

Brown & Itami model:


Making sense
Visual array
Relative relief
Spatial diversity
Relief contrast
3-D space
Height contrast
Internal variety

Brown, Keane and Kaplan [1986] tested this model by comparing the preferences obtained for scenes with those predicted by the Brown & Itami model. The correlation of 0.61 is significant at p< 0.001.

A further analysis was undertaken by grouping scenes using factor analysis; four groupings were obtained [Table 8.3]. Comparison of the predicted average values and preference ratings indicated identical rankings for the two procedures [5-point scale].

Table 4   Relationship between predicted values & preference ratings

    Mean Rank Mean Rank
Manicured landscapes
Mostly vegetation

Source: Brown, Keane & Kaplan, 1986

According to the authors, the results provide support and encouragement for further work. The higher preference values occurred for smooth-textured grassy areas, suggesting that coherence is more important than indicated by the model. Similarly, low preference values occurred in relatively barren scenes, suggesting the importance of complexity.

Gimblett, Itami & Fitzgibbon [1985] asked respondents to rate photographs on the basis of the Kaplans’ dimension of mystery using a 5-point scale. Analysis found a high degree of agreement regarding mystery in the landscape and analysis of the photographs identified five attributes that were associated with mystery [Table 5].

Table 5   Physical Attributes of Mystery

Mean ratings & mystery class 1.0 - 2.0
2.0 - 3.9
4.0 - 4.5
partial to full
Distance of view
Spatial definition
partially enclosed
Physical assessibility
defined path
Radiant forest
  forest illumination

Source: Gimblett, Itami & Fitzgibbon [1985]

The five physical attributes were defined as follows:

  • Screening: degree to which views of the larger landscape are visually obstructed or obscured
  • Distance of view: measured from viewer to nearest forest stand; as distance increases, mystery decreases
  • Spatial definition: degree to which the landscape elements surround the observer
  • Physical accessibility: apparent means of moving through or into the landscape as a result of finely textured surfaces in the foreground; provides way of exploring landscape to gain more information
  • Radiant forests are special cases in wooded areas where the immediate foreground is in shade and an area further in the scene is brightly lit. These are consistently ranked high for mystery.

Gobster & Chenoweth [1989] analysed the physical, artistic and psychological variables of landscapes and found that all three aspects could explain preferences. The ten psychological descriptors included mystery, harmony, legibility, awe and pleasantness. They also found that the three variables were interrelated within a definable structure. A conceptual interrelatedness was also found between descriptor variables with the artistic and psychological dimensions defining separate constructs relating to the compositional and affective-informational meanings. Multi-dimensional scaling indicated that the psychological descriptors yielded the highest multiple correlation of R = 0.84 [< 0.0005], significantly higher than that for the physical descriptors [r = 0.67, p < 0.05] or artistic descriptors [R = 0.69, p < 0.05].

They concluded:

“These findings should be of interest to those concerned with theory and application in landscape research. Aesthetic theories based solely on formal-artistic, bioevolutionary and other singular sets of properties (i.e. physical-ecological, psychological-affective) etc may not do justice to the richness of human aesthetic response to landscapes. To build an aesthetic theory of landscapes, investigators need to broaden their understanding of the multidimensional nature of aesthetic preferences.” [my emphasis]

In Gregory & Davis [1993], the positive factors [trees, tree trunks and water depth] can be considered as contributing to the legibility and coherence of a riverscape, while the negative factors [water colour, bank channelisation, channel sinuosity and debris in the river] may be considered as contributing to the complexity and mystery of the scene. These are my interpretations; the authors did not assess the riverscapes in informational terms. Water colour, bank stability and water depth together accounted for nearly 90% of the variation in the riverscape preferences.

Thomas Herzog undertook a series of studies in the 1980s to explain and assess the validity of the Kaplans’ information processing model.

In Herzog’s [1984] study of field and forest environments, moderate correlations [0.45 to 0.55] were obtained for the three predictors of the unconcealed vantage point dimension: identifiability [i.e. familiarity], coherence and spaciousness. These help one organise and make sense of a setting in Kaplans’ terms. Herzog comments that “their prominence as predictors suggests that when one is out in the open, there is a premium on being able to figure out where one is and where one could get to quickly." In the large trees category, high ratings were obtained for the making-sense [i.e. identifiability, coherence, texture] and involvement [i.e. mystery] properties, which supports the Kaplans’ contention that scenes high in both of these properties will be most preferred. Herzog [1985] used the same predictor variables to rate waterscapes [Figure 6] and found:

  • Spaciousness was best shown in large water bodies; these also showed highest texture and coherence but lowest complexity and mystery - these water bodies lack interest and are easy to make sense of;
  • By contrast the other water bodies are more interesting, being high in mystery and complexity yet being reasonably coherent;
  • They thus reward immediate involvement yet hold out promise of more;
  • The distinguishing features of [1] mountain waterscapes are their low textures which suggest that they are difficult to navigate; [2] low spaciousness of swampy areas; [3] identifiability of rivers, lakes & ponds; [4] large bodies of water have the most distinguishing features.


    Source: Herzog [1985]
    Figure 6   Rating of Waterscapes by Variables

    Waterscapes high in spaciousness, coherence and mystery but low in texture [e.g. uneven land] were preferred. Inter-correlations with preference were: spaciousness 0.42 [p < 0.01], coherence 0.33 [p < 0.01], mystery 0.09, texture -0.15 [Ibid, 235]. Those that are at least moderately high in making sense [understanding] and involvement [exploration] were preferred. The content of the water is also important; rushing water is preferred over stagnant creeks. Herzog found the information approach useful in accounting for waterscape preferences.

    Herzog [1987] examined mountainous scenes using the same six predictor variables and preference as the criterion variable [Figure.7]. He found:

    • deserts are low in spaciousness, [the predictor is a feeling of spaciousness offered by the scene] but are only moderate in other ratings
    • snowy mountains are high in spaciousness but are of low complexity while smaller mountains are also high in spaciousness and identifiability
    • narrow canyons have the most extreme profile being low in spaciousness, texture and identifiability but very high in mystery. Spacious canyons [e.g. Grand Canyon] are high in spaciousness, coherence and complexity.


    Source: Herzog [1987]
    Figure 7 Rating of Mountainous Scenes by Variables

    Intercorrelations with preference were: identifiability 0.61 [p < 0.00], spaciousness 0.32 [p < 0.01], texture 0.22 [p = 0.06], mystery 0.13 [p = 0.29]. While the mountain categories are reasonably high on spaciousness, the two canyons differ markedly on this variable. The difference in identifiability between the mountain scenes is likely to be due to the familiarity of small ranges to the participants. The lower rating of texture for small mountains reflects their less smooth, more rugged appearance of the snowy mountains, in which snow and clouds tend to obscure their true ruggedness. As texture reflects the affordance of locomotion the results suggest that this is not validly measured by texture.

    Again Herzog found the informational approach useful in accounting for natural landscape preferences and supported the approach of examining both content and cognitive processes in the evaluation of these preferences. The “pattern of significant variables changes substantially when content categories are included." A positive predictor of preference is identifiability [i.e. familiarity] that gives “eloquent testimony to the strong cognitive need to make sense of the environment in such settings.”

    The basic predictor variables as established by the Kaplans were developed in other studies. Strumse [1994b] applied them, together with perception-based variables [e.g. openness, smoothness, ease of locomotion] in western Norway, and found the informational variables were the most effective predictors of preference [r2 of 0.66]. Ulrich [1977] developed focality [i.e. a focal point], as an extension of coherence, ground textures as a factor in complexity, and depth, or a sense of space, as an element in exploration and legibility. Whitmore [1995] applied the basic predictor variables to a canyon landscape, describing water, vegetation and landforms in informational terms.

    The Kaplans’ theory has been subjected to a range of studies and they all provide support for its elements. There would appear, however, to be a fair degree of interpretation required of the application of these four predictor variables in the landscapes studied. The nebulousness of the concepts involved suggests that they are still evolving and this is likely to continue for some time.

    The predominance of photo ranking as the main instrument used in the studies is worth noting. The nine studies by the Kaplans and their colleague, Herzog, contributes to this dominance. Out of the total of 227 studies only 29% used photo ranking but 84% of the information processing studies used it.

    Stephen Kaplan acknowledges that his approach is an evolutionary view based on habitat theory, with human preferences deriving from the adaptive value offered by particular settings [Kaplan, 1987]. Preferences were regarded by Kaplan as:

    “an intuitive guide to behavior, an inclination to make choices that would lead the individual away from inappropriate environments and towards desirable ones.”

    He stated:

    “The central assumption of an evolutionary perspective on preference is that preference plays an adaptive role; that is, it is an aid to the survival of the individual.” [1982].

    Every aspect of preference should provide some “discoverable benefit or payoff” [Ibid]. Deriving environmental preference occurs very rapidly and unconsciously. It is:

    “the outcome of what must be an incredibly rapid set of cognitive processes which integrate such considerations as safety, access and the opportunity to learning into a single affective judgement.”

    Kaplan considered that the character of predictor variables and the nature of preference responses support an evolutionary interpretation. In support, he cited the preferences for savanna [Balling and Falk, 1982], the similarity of landscaped parks to savanna [Orians, 1986] and the prospect-refuge theory of Appleton [1975].  An evolutionary analysis, Kaplan asserted, achieves a number of objectives, it:

    • Indicates the importance of preference
    • Provides an expectation of underlying commonality in preferences across individuals
    • Suggests that preference research has a substantial theoretical interest
    • Identifies variables likely to be effective in predicting preference [1982, 187]

    An evolutionary viewpoint led Kaplan to conclude that:

    “Aesthetic reactions reflect neither a casual nor a trivial aspect of the human makeup. Aesthetics is not the reflection of a whim that people exercise when they are not otherwise occupied. Rather, such reactions appear to constitute a guide to human behaviour that has far-reaching consequences.” [Kaplan, S, 1987]

    Kaplan went on to state that organising workspace, arranging one's home, avoiding certain directions and approaching others may reflect factors such as coherence, legibility, mystery and complexity. He concluded that there is clearly more to aesthetics than optimal complexity and that the "acquisition of new information and its comprehension (are) central themes underlying the preference process."

    Zube summarised the Kaplans’ approach thus [1984]:

    "The Kaplans propose that long term survival of the human species was dependent upon development of cognitive information processing skills which in turn led to preferences for landscapes that made sense to the observer. In other words, landscapes were preferred that could be comprehended, where information could be obtained relatively easily and in a non-threatening manner that provided opportunity for involvement, and that conveyed the prospect of additional information. According to this framework, landscapes that are preferred are coherent, legible, complex, and mysterious."

    Balling and Falk summarised Stephen Kaplan’s contribution [1982]:

    “Taking an evolutionary perspective, S. Kaplan has asserted that the long-term survival of the extremely knowledge-dependent human species required that people should actually like to obtain information about landscapes, and that they should be able to process certain kinds of environmental information very efficiently.”

    Bourassa notes that the information processing theory emphasises “only some of the biological bases for aesthetics, not to mention the fact that it ignores cultural and personal modes of aesthetic experience” [1991].

    Go to top

    Tripartite Paradigm of Aesthetics - Stephen Bourassa

    Stephen Bourassa, now at the Department of Urban and Regional Planning, University of Sydney, worked for several years in addressing the biological, cultural and personal attributes of landscape perception. He published several papers andlater consolidated them in The Aesthetics of Landscape [1991]. The following sums up his quest:

    “If both biology and culture serve as distinct bases for aesthetic behavior, then it is necessary to go beyond both biological and cultural determinism toward a theory which would fully embrace both biological and cultural factors. It is also necessary to consider the role of personal idiosyncrasies and particularly personal creativity...” [Bourassa, 1991].

    Bourassa drew on the work of the Russian psychologist, Vygotsky. Vygotsky was regarded as a non-person in Stalinist Russia and his ideas have been slow to appear in English. He sought to accommodate both the biological and cultural aspects of behaviour. He focussed on the process of development rather than its product and, in so doing, was able to provide explanations of behaviour rather than mere descriptions. Vygotsky’s tripartite development approach is summarised in Figure 8, together with the three modes of aesthetic experience suggested by Dewey’s theory of aesthetics.


    Source: Bourassa, 1991
    Figure 8    Vygotsky’s Development Paradigm + Dewey’s Modes of Aesthetic Experience

    Bourassa is cautious about paralleling Dewey’s modes of aesthetic experience with Vygotsky’s theory, but noted that the eminent 18th century Scottish philosopher, David Hume, also suggested a tripartite basis for aesthetics. In his book, Treatise of Human Nature, Hume wrote: “beauty is such an order and construction of parts, as either by the primary constitution of our nature, by custom, or by caprice is fitted to give a pleasure and satisfaction to the soul.” [Quoted in Bourassa, 1991]. Hume’s categories are remarkably similar to Vygotsky and Dewey.

    Bourassa questioned whether the aesthetic experience is separate for the biological and cultural modes or whether they are inextricably intertwined. Based on work of the neurophysiologist, P.D. MacLean, Bourassa believed there are dual modes of perception. The neurophysiological research suggests that:

    “instinctual and emotional responses to landscape could occur separately from rational and cognitive responses. In other words, there could be separate innate and learned responses to landscape.” [Ibid, 59]

    Similarly he quoted Izard: “although emotion and cognition are in large measure interdependent, another body of evidence suggests as well that emotion processes and cognitive processes have a significant degree of independence.” While cognitive psychology assumes that feeling follows cognition, Bourassa also quoted Zajonc’s [1980] argument that affect is pre-cognitive, citing a lack of evidence for the post-cognitive view [see also Kaplan, 1987]. Bourassa cited experiments that have demonstrated preferences for stimuli, even in the absence of any cognitive knowledge of these stimuli. Bourassa urged caution on the issue of pre-cognitive affect and summarised the position thus:

    “The research findings ... suggest that:

    • There are dual perceptual systems involving both the uniquely human and the more primitive parts of the brain
    • The more primitive parts of the brain function on the basis of emotion rather than cognition
    • The primitive brain can respond to stimuli in the absence of cognitive awareness of those stimuli
    • Consequently, affective response to stimuli may under some circumstances occur separately from cognitive knowledge.”

    Based on this, Bourassa concluded that ‘biological’ responses to landscape could occur separately to ‘cultural’ responses. Based on work by Meyer [1979], he then argued that the three levels [biological, cultural and personal] require respectively aesthetic laws, rules and strategies.

    At the biological level, he reviewed Appleton’s prospect-refuge theory, habitat theory and information processing theory. At the cultural level, he reviewed Costonis’ cultural-stability-identity theory of aesthetics in which groups seek to perpetuate the symbolic landscape as a means of self-preservation. Finally, for the personal level, he reviewed theories of creativity and its role in landscape perception.

    Having established biological, cultural and personal dimensions of landscape perception, Bourassa then sought to demonstrate its application. He noted, for example, that the preference found for natural scenes over urban ones could be explained by his tripartite paradigm; natural landscapes are experienced more in the biological mode while urban landscapes are experienced more in the cultural mode. He also considered that the formalist, objectivist approaches involving quantitative measurement of landscapes could only be applied to the biologically based preferences:

    “Outside of that realm, cultural and personal values must also be considered and landscape aesthetics must be viewed in terms of the experiential interaction of the perceiver and the landscape.”

    On this basis, he was critical of the method by Shafer et al [1969] of deriving regression equations from analyses of landscape photographs, a “kind of gross empiricism [which] can often lead to spurious results.”

    Although Bourassa has provided a service to landscape interests by constructing an integrated framework within which to consider the biological, cultural and personal dimensions of landscape preferences, it is questionable whether it amounts to little more than a framework or paradigm.

    While he initially referred to the need for a theory [p 49] and to his “tripartite theory of aesthetics” [p 64], he subsequently referred to it as a “tripartite framework” [p 66] and a “tripartite paradigm” [p 120]. However, in his final chapter on postmodernism [the relevance of which is unclear], he reverted to referring to “the aesthetic theory presented in this book” and the “aesthetic theory developed in Chapters 1 to 6” [p 133]. It must therefore be assumed that while Bourassa had doubts himself as to whether he had established a theory, on balance he felt that he had.

    Based on the Shorter Oxford English Dictionary definition of theory as “a systematic statement of rules or principles or a scheme or system of ideas as an explanation of facts or phenomena”, the benefit of the doubt should be given and Bourassa’s contribution regarded as a theoretical framework. However, despite the critiques he offered of various existing techniques, the application of the framework to the determination of landscape quality is not clear. Nor are there clear ways by which it could be tested or applied in a predictive manner. Nevertheless, it does provide a comprehensive integrated framework covering the three dimensions which can be used to inform further analysis and to assess the results of studies.

    In a review of Bourassa’s The Aesthetics of Landscape, Seamon [1993] was critical of Bourassa on a number of counts, including a “bias against a formalist approach to landscape”, an ignorance of phenomenological research which is supportive of landscape contributing to the aesthetic experience, and his reduction of the aesthetic experience to “the three rather standard ... dimensions of biology, culture, and individual.” Overall, he considered Bourassa’s theory “provides little understanding of the powerful feelings that landscape, place, and environment can evoke...”

    Go to top

    Phillip Dearden’s Pyramid of Influences

    A model postulated by Phillip Dearden [1989] of the University of Victoria, British Columbia has close parallels with Bourassa’s tripartite paradigm [Figure 9].


    Source: Dearden, 1989
    Figure 9 Dearden’s Hierarchy of Societal Landscape Preferences

    Dearden noted [1989] that the hierarchy is not intended to imply the relative importance of the variables but rather recognises that each variable is present in influencing landscape preferences. The emphasis of the hierarchy is to reflect the potential degree of social consensus related to each variable. Innate factors deriving from human evolutionary history are common for all people; cultural factors are common for a particular society, while factors such as familiarity and socio-economic and demographic factors are far more related to particular individuals in time and space.

    Based on this model, Dearden suggests that the techniques for landscape assessment need to relate to the degree of individual differences. Techniques which are landscape based [objectivist] are appropriate in assessing innate and cultural factors, but techniques which provide for greater probing of individual perceptions [subjectivist] are appropriate for assessing individual influences.

    Go to top


    “Diverse Theoretical Origins”

    Having reviewed the key theoretical constructs, we return to the assertion by Sell, Taylor and Zube that landscape perception research is characterised by a “scattering of diverse theoretical origins.” [1984]. In their original work, Zube et al found a diversity of theoretical backgrounds to the literature: art theory, ecological concepts, stimulus-response behaviourism, signal detection theory, adaptational theories such as ‘optimum stimulus level’, ‘prospect-refuge’ and ‘information processing’; personal construct theory, behaviour-setting theory, phenomenology and transactional theory [1982].

    Sell, Taylor & Zube [1984] grouped these theoretical sources by the paradigms they identified. In many cases the theoretical origins are implicit and assumed rather than explicitly defined in the studies. Zube  [1984] described them as “theories and concepts that are embedded, but not always explicit, in much of the work.”

    Zube went on to describe the theoretical origins according to the disciplines involved in landscape assessment:

    Disciplines Theoretical origins
    Planners, landscape architects, natural resource managers
    Principles of visual aesthetics and landscape design, ecological theory and biological resource management
    Behavioural scientists
    Signal detection, stimulus-response, arousal, adaptation level and information processing
    Humanists and cultural geographers
    Sense of place, transactionalism, historicism, phenomenology

    While several of these constitute theoretical constructs, others are simply in the form of principles or “rules of thumb” developed by professionals in a discipline. Zube [1984] drew on work by Moore et al [1982] in proposing a four level structure of theory:

    Level 1
    Theoretical orientations or general theories representing broad concepts that serve as heuristics in orienting ways to look at phenomena and to identify lines of research
    Level 2
    Frameworks representing relationships among existing findings that provide a conceptual and systematic organisation to data about phenomena
    Level 3
    Conceptual models which provide descriptions of variables and of relationships among variables but not necessarily explanations of phenomena within a larger theoretical context
    Level 4
    Conceptual models which provide descriptions of variables and of relationships among variables but not necessarily explanations of phenomena within a larger theoretical context

    Zube suggests that most of the work has been in levels 3 and 4, which seems a rather generous assessment. Assuming that levels 3 and 4 reflect greater levels of specificity I suggest that habitat theory is at level 1, while information processing and prospect-refuge theories are level 2. Bourassa’s tripartite paradigm and Dearden’s hierarchy of preferences appear also to be level 2.

    Within the social sciences, three main approaches to theory generation have been suggested [Sancar, 1985]:

    Universalistic Abstracts, formalises and generalises relations using a hypothetico-deductive approach
    Situational Generates contextually relevant information for planning and management in specific settings
    Integrative Through induction, generates grounded theory which is based on the premise that the adequacy of a theory cannot be divorced from the process by which it was generated

    Sancar considered that Zube et al’s expert and psychophysical paradigms are situational, while their cognitive and experiential paradigms are of the universalistic type. She considered that none of the paradigms may be associated with the integrative approach. She considered the “need for an integrative approach to fill the theoretical void in landscape aesthetics research."

    As much research seeks to verify a preconceived theory but the real issue is the theoretical void that exists, she suggested the real need is for theory generation. She proposes the “grounded theory” approach, which is “based on the premise that the adequacy of a theory cannot be divorced from the process by which it is generated.” [Ibid] This may be achieved through comparative analysis, use of quantitative and qualitative data and secondary analysis of substantive data. In particular, the characteristics of the theory would derive from those cases where the following criteria are achieved:

    Internal validity Conditions are reliably represented
    External validity Conditions typify those found in other situations
    Reflexivity New concepts are generated by comparing information obtained through different methods
    Translatability Consensus is promoted with conflicting frames of reference

    These criteria derive from work by Dunn and Swierczek [1977] and are used by Sancar to develop the procedure for what she terms “a reflective-dialectical strategy of inquiry and choice” emphasising the generation of theory rather than the testing of theory.

    Carlson [1993] distinguishes between explanatory theory, the kind used in science to explain, predict and control, and that which he terms justificatory theory, with its origins in philosophy. Justificatory theory:

    “concentrates on our ideas or concept of things, indicates the reasons why these ideas and concepts are as they are, and thereby aids in justifying our views about things.”

    He suggests, that although writers have noted the theoretical vacuum in landscape studies, it is the justificatory form rather than the explanatory form that should be sought. In contrast to the explanatory form, a justificatory theory seeks to explain why the subject [e.g. landscape quality] is important in our lives. Commenting on Bourassa’s approach, he considers that, although it is “rich in orientational, organizational, and explanatory power, [it is] poor in justificatory power.”

    He believes that justificatory theory is not imposed but rather grows out of a field, being:

    “the result of a lifetime of experience in and appreciation of the landscape, together with deep and reflective thought about the nature and the meaning of such experience and appreciation.”

    The influence of Carlson’s ideas has yet to be seen in landscape research.

    Clearly a robust theory of landscape which provides an all-encompassing framework with which to understand and to predict landscape preferences, does not currently exist. At present there is a range of theories that offer explanations of aspects of landscape preferences but which fall short of a definitive explanation.

    Of the theories available, the Kaplans' information processing theory appears the most supportable, based on the range of studies that have assessed its validity and explored the dimensions of the factors involved.

    Appleton’s prospect-refuge theory has intuitive appeal but the studies undertaken fail to provide conclusive support, if anything tending to indicate its shortcomings and areas in which the evidence is contrary to the theory. Some of his elements have parallels with the dimensions of the Kaplans’ information processing [e.g. prospect and legibility, refuge and mystery], although it is acknowledged that each area is coming from very different intellectual positions.

    Ulrich’s affective theory has good support from studies but, like habitat theory, its usefulness in understanding and predicting landscape preferences is limited. Rather it focuses on the positive effect that landscape can play on emotional states of well being.

    While the Kaplans’ theory offers the most comprehensive explanation of landscape preferences, it is not a theory that is readily applicable in a field situation to evaluate landscape. By contrast, the appeal of Appleton’s and Orian’s theories is that they offer explanations that can be readily applied in the field.

    If the mark of solid theory is in its use in applications, then none of the theories currently available provide a useable framework for the evaluation of landscape in a field situation. While they can offer tantalising glimpses of understanding, they fall well short of comprehensively enabling the evaluation of landscapes.

    The conclusion of Gobster and Chenoweth [1989] is confirmed, existing theories based on artistic, bioevolutionary or other properties fail to capture the “richness of human aesthetic response to landscape”. They suggest the need for researchers to “broaden their understanding of the multidimensional nature of aesthetic preferences.”

    Go to top


© 2009-14 Scenic Solutions   Privacy   Terms of Service   Site Map   Contact us