Concept identification and environmental perception: Classification and evaluation in visual landscape assessment.

Abstract

This project was designed to extend the principles of natural categorization to the classification of landscapes for visual quality assessment. In the first study, 20 lay people named and outlined distinct geographic units on USGS topographic maps. Six of the units identified were selected for further study on the basis of ratings of overall environmental quality, familiarity, and naturalness. Consensual names and boundaries of the units were determined. In the second study, the same 20 subjects rated 15 scenes from each of the 6 units for representativeness (typicality) and visual quality. The ratings of representativeness and visual quality were highly reliable, with coefficients ranging from.98 to.84. The correlations between representativeness and visual quality were variable. The correlations were positive for the two high environmental quality units (r =.78 and r =.83, p<.05). Representativeness and visual quality were positively related for one of the two moderate quality environments (r =.53, p<.05). In the two low quality environments, the correlations were non-significant, but for one of these units there was a negative trend (r = -.45), and this relationship was significantly different from the other five correlations. Overall, these results suggest that the principles of natural categorization are active in the conceptual analysis of environments, judgements of representativeness and visual quality are reliable, and judgements of representativeness and visual quality are not the same. Judgements of representativeness can be used in resource decision making to provide reliable information about what is characteristic of an environment and to determine how development proposals relate to the existing character of an area. In addition, the principles of natural categorization are used in connectionist models to explain how humans identify objects and develop concepts. The principles of natural categorization are active in environmental perception, but it remains to be seen whether the connectionist approach can provide adequate models of environmental perception. This research provides a method which can be used to study how environmental perception relates to natural categorization.