Three issues related to the prediction of aesthetic response of natural landscapes were investigated. First, information regarding the degree of correspondence between two conceptually different yet commonly used criteria of aesthetic response--ratings of scenic quality and preference ratings--was sought. Second, the relative efficiency of and interrelationships between predictor variables stemming from different philosophical paradigms of landscape aesthetics was of interest. Examination of such a variety of predictors towards the same criteria utilizing the same settings as stimuli had not been previously researched. Direct comparison of types to one another, and in combination as predictors, would indicate both whether different approaches were measuring similar aesthetic response variance, and in what ways they differed. Third, the extent to which a motivational choice model based in expectancy theory could predict environmental preference was of interest. This model represented an aesthetic predictor in terms of environmental utility, i.e., meaningfulness within the context of potential activity, and was thus a departure from traditional predictors based on design elements and the arrangement of physical features.

Data were gathered from a total 354 subjects responding to 60 different natural landscape scenes (color slides) from a wide variety of United States' biomes. Results indicated that the two aesthetic criteria were nearly identical, both in relation to one another (r=.98) and through their correlate patterns to 33 predictor variables. Predictor variables from three paradigms: the psychophysical (physical features of the environment), the cognitive (transactional variables involving interpretive patterning of physical variables), and the experiential (environmental utility in terms of potential for activity) were all highly effective. Multiple regression equations for specific types had predicted R-Squares ranging from .47 to .84. In turn, detailed analyses of the transactional and utility variables via multiple regression (using the physical variables as predictors) indicated they could be defined by these managerially controllable terms. Finally, the environmental utility variable was examined in more detail through a variety of expectancy models. Of major interest was that environmental familiarity was a strong moderator of the utility effect, with highly familiar settings yielding more accurate prediction than unfamiliar settings. A number of managerial implications and suggestions for follow-up research are made.