Elsevier

Landscape and Urban Planning

Volume 65, Issue 4, 15 November 2003, Pages 261-277
Landscape and Urban Planning

Subjective responses to simulated and real environments: a comparison

https://doi.org/10.1016/S0169-2046(03)00070-7Get rights and content

Abstract

In order to assess the validity of computer-generated environment simulations, an empirical field study was conducted. First a computer model of a real urban park environment was developed and used to produce both daylight and night-time animations of a 3 min walk into and through the park. The level of visual detail is high with all trees, buildings and hard surfaces correctly textured. Moving vehicles are also included. Sounds recorded on-site along the selected path were dubbed onto the animations and recorded on videotape. Then an elaborated questionnaire was constructed which measures respondents’ cognitive and affective reactions to the presented environment, including impressions of the area, content retention and comprehension, and their evaluation of the simulations’ realism. Four groups of participants saw the animations and were also taken for a walk in the real environment, either by day or night; for half of them the order of simulation and reality was reversed. The results show that even detailed and time-consuming computer simulations do not necessarily generate the same responses as the corresponding real environment. However, differences between day and night conditions are mostly the same in the simulated as in the real environment, and the realism ratings of the viewers were generally encouraging. The findings elucidate where further development and evaluation are warranted.

Introduction

The critical assessment of environmental simulations in relation to the depicted reality is important because the simulation of physical environments using computer graphics is becoming a commonplace occurrence. Meanwhile, the psychological validity of such presentations has received far less attention. The capacity to generate highly realistic environmental simulations has developed with increasing computer power and sophistication in rendering algorithms. Landscape, environmental and urban researchers have begun to adopt the new technology for the purpose of design communication (e.g. Clipson, 1993, Decker, 1994, Levy, 1999), resource management (e.g. Daniel, 1992) and research into human responses to environments (e.g. Bishop and Rohrmann, 1995, Rohrmann et al., 2000). Recently, Pietsch (2000) has reviewed the several roles of computer visualisation in urban design.

Environmental simulation is particularly important for two purposes: the communication of the impacts of planning and design in environments which do not yet exist, and controlled experiments into perceptions and evaluations of environmental change. Many different surrogates for direct experience of landscapes and built environments have been used in the past (Zube et al., 1987, Sheppard, 1989) and for many purposes these simulations may appropriately be highly abstracted (Pietsch, 2000). The use of simulated environments in the study of human perception, on the other hand, requires high levels of sensory realism (Daniel and Meitner, 2000). As our longer term objective is to use virtual environments to understand human perception and behavior better—not just to support planning decisions—sensory realism is the focus of this paper.

The literature clearly indicates that the greater the degree of realism in the simulation the more effective, in this context, it becomes.

For centuries it appears to have been assumed that a drawing—is a drawing—is a drawing, and that it probably means the same thing to all who view it. The evidence … suggests that the most realistic simulations, those that have the greatest similitude with the landscapes they represent, provide the most valid and reliable responses. (Zube et al., 1987, p. 76)

Since Zube et al. reviewed the range of options in environmental simulation, and drew this conclusion, computer graphics have provided increasingly accessible platforms for the achievement of similitude in simulation. Stamps (1990) used meta-analysis to show that photographs are a good surrogate for direct experience in the evaluation of scenic preference. A number of studies have addressed the issue of the validity of computer simulations in comparison to photographs (e.g. Bishop and Leahy, 1989, Daniel and Meitner, 1997, Oh, 1994, Bergen et al., 1995). While these studies have generally supported the conclusion of Zube et al. (1987) that a good simulation can be used as a surrogate for a direct view of the environment, they are not necessarily sufficient since:

  • these studies have primarily been based on static imagery (cf. Hetherington et al., 1993);

  • none has undertaken a direct comparison between computer simulations and first-hand environmental experience;

  • the majority of comparative studies have restricted themselves to comparisons of preference ratings and have not addressed other important aspects of environmental response (Bishop and Hull, 1991, Bishop and Rohrmann, 1995). Yet a comprehensive understanding of the multiple aspects of subjective appraisal of environmental quality is important for understanding of human–environment interactions and, we believe, effective environmental planning (Rohrmann, 1988).

Consequently, research should address these deficiencies
  • by working with animations rather than static images;

  • by taking subjects to the actual site rather than relying on photographs as ‘reality’;

  • by comparing a range of perceptual responses in the real and simulated environments.

We also note that an interactive experience (Bishop and Dave, 2001) is superior to viewing an animation. However, at this stage of computer development the realism of a real-time simulation of an exterior environment is inferior to that created by frame-by-frame rendering and playback as an animation.

Perceived environmental quality (Craik and Zube, 1982) is a subjective transformation of ‘objective’ features of the environment according to individual experiences and preferences. Environmental psychologists have identified several types of cognitive, affective and conative responses (e.g. Gaerling and Golledge, 1993, Gifford, 1997, Kaplan and Kaplan, 1982, Nasar, 1988, Stokols, 1988, Ulrich, 1986):

  • identification (for objects and structures, according to existing knowledge);

  • orientation (depending on the “legibility” and the novelty of the environment);

  • encoding (storage of the perceived environment in memory);

  • aesthetic evaluation (perceived beauty and congruity according to individual standards);

  • personal liking (subjective pleasantness, familiarity, historical and symbolic value);

  • adaptation and safety (behavioural intentions to conform with physical reality);

  • manipulation (modifying an environment for personal utilisation).

For an environmental simulation to be considered valid it should evoke an equivalent set of responses, in each such category, as would direct experience of the same environment. The two sets cannot be expected to be identical because it is impossible to match the richness and complexity of reality. However, the more similar they are the greater the faith we can place in environment simulation as a tool for, e.g. evaluating alternatives in environmental design, for providing stimuli for perception research, or for using simulations in environmental hazard education. The search for generality can be extended to the presentation of different weather and light conditions, i.e. the validity of a sunny versus ‘grey’ situation or day versus night simulations. Ideally, all should be tested specifically, but according to the same set of responses.

Fig. 1 shows the conceptual framework developed for a series of studies on environment simulations (cf. Rohrmann et al., 2000). It integrates the major factors to be considered in an evaluation of computer simulations and identifies the main influences on response variables. What respondents see in a presentation results from a combination of the characteristics of the objective environment (the box labelled “real world”) and the chosen presentation means (the triangle labelled “mode”), i.e. various forms of experience of the environment, including computer simulations, video recordings and site visits. These in combination determine the “features of the presentation”. In Fig. 1, all variables to be measured as viewer’s responses are shown as circles (dotted: appraisal of the represented environment; solid: reactions to the presentation). The core variable, perceived realism (double-circled) refers to an evaluating judgement in which the simulation, in relation to the viewer’s beliefs about the reality, is assessed. The figure also makes clear that judgements about the environment itself and about its depiction—which is dependent on the available representation means—are likely to interact (double-headed arrow, indicating mutual causality). Real environments as well as simulated ones are always experienced within a subjective context of cognitive and evaluative factors.

The principal aim of this study is to clarify the validity of carefully prepared computer simulations for representing urban environments under day-time and night conditions. In particular, the following research questions were posed:

  • Are evoked cognitive and affective responses to the simulations similar to those when exposed to reality?

  • Is the level of information recalled similar for real and for simulated environments?

  • What level of realism rating will people attribute to computer simulations and what features contribute to the realism rating?

  • Are night-time simulations more or less valid in their induced perceptions than the daylight equivalents?

These questions cover a range of response types (above). However, as the project focuses on the perception and evaluation of environments, the cognitive impacts (e.g. comprehension and retention) and affective responses (e.g. impression and likeability) are of prevalent interest. Both aspects are crucial for the perceived realism of simulations. Given the computer resources available at the time of the experiments, adaptation or manipulation responses could not be studied. These require an interactive environment as discussed in Bishop et al. (2001).

Landscape and urban designers are among the major potential users of computer simulation as a communication tool. Their operational environment commonly includes public places near major buildings. Landscape perception research is often focused on the natural environment but also includes many urban studies (Kaplan and Kaplan, 1982, Smardon, 1988, Herzog and Chernick, 2000). For generality, we therefore chose our study area in the environs of a suburban civic centre. It includes a major transport route, shops, civic buildings and parkland. Such an environment is heavily used in both the day-time and after night-fall. A representative simulation must therefore include presentations of both daylight and artificial lighting conditions. This is also relevant for impressions such as safety versus threat (cf. Nasar et al., 1993, Painter, 1996), as fear associations are considered more likely for the night situation.

The current study is the third component of a larger project on the validity of environment simulations. In study (1), several variations in a simulation of the environment mentioned above were investigated (Rohrmann and Bishop, 2002). In study (2), a computer simulation and a video recording of that environment were compared (Palmer, 1998, Rohrmann et al., 2000). All three studies used simulations of the same area.

Section snippets

Design of the experiment

In order to address these questions, a combined field/laboratory experiment was designed. It was based on a 3 min walk through the chosen environmental setting. Participants were exposed to both the actual and the simulated environment (half of them in reverse order). Thus, “presentation mode” (cf. Fig. 1) means either a site visit or a computer simulation. Different groups were employed for the “day” and the “night” condition. The resulting mixture of between and within groups design is

Overview

Responses to the real and simulated environments were compared in terms of the variables listed in Table 2. Subjective assessments of simulation realism and ability to match specific features were analysed based on both direct and comparative ratings. It was considered important to determine not only the degree to which mean responses compared but also how scores were distributed between respondents. That is, an effective simulation should generate a similar profile of response as will the

The validity of responses to simulated environments

The results suggest that detailed and often laborious computer simulations can provide valid outcomes for the main aspects of environmental perception but, from several points of view, do not generate the same responses as the corresponding real environment. Using a simulated environment, the appreciation of the study area is less positive, and retention is less detailed. However, differences between day and night conditions are mostly the same in the simulated as in the real environment.

Of

Acknowledgements

This research was supported by a grant from the Australian Research Council (ARC). We would like to thank our research assistants, Dean Lusher and Caitlin McLeod, for their competent help with collecting and analysing the data. The two reviewers of our initial submission made many good points and have helped considerably with the clarity of this paper.

References (50)

  • K. Painter

    The influence of street lighting improvements on crime, fear and pedestrian use, after dark

    Landsc. Urban Plann.

    (1996)
  • M.R. Patsfall et al.

    The prediction of scenic beauty from landscape content and composition

    J. Environ. Psychol.

    (1984)
  • B. Rohrmann et al.

    Subjective responses to computer simulations of urban environments

    J. Environ. Psychol.

    (2002)
  • M.J. Scott et al.

    Picture or place? A multiple sorting study of landscape

    J. Environ. Psychol.

    (1997)
  • R.C. Smardon

    Perception and asthetics of the urban environment: review of the role of vegetation

    Landsc. Urban Plann.

    (1988)
  • R.S. Ulrich

    Human responses to vegetation and landscapes

    Landsc. Urban Plann.

    (1986)
  • H.A.H.C. van Veen et al.

    Navigating through a virtual city: using virtual reality technology to study human action and perception

    Future Gen. Comp. Syst.

    (1998)
  • E.H. Zube et al.

    Landscape perception: research, application and theory

    Landsc. Plann.

    (1982)
  • Bishop, I.D., Dave, B., 2001. Beyond the moving camera: systems development for interactive immersive exploration of...
  • Bishop, I.D., Leahy, P.N.A., 1989. Assessing the visual impact of development proposals: The validity of computer...
  • Bishop, I.D., Rohrmann, B., 1995. Visualisation tools: their role in understanding human responses to the urban...
  • I.D. Bishop et al.

    Extending the geographic information base into the third dimension for use in the urban environment

    J. Urban Region. Inf. Syst. Assoc.

    (1995)
  • I.D. Bishop et al.

    Experimental approaches to perception response in virtual worlds

    Landsc. Urban Plann.

    (2001)
  • Clipson, C., 1993. Simulation for planning and design: a review of strategy and technique. In: Marans, R.W., Stokols,...
  • Craik, K.H., Zube, E.H., 1982. Perceived Environmental Quality. Plenum Press, New...
  • Cited by (0)

    View full text