An investigation of computer-generated architectural representations on design decision-tasks

The use of computer-generated visual simulations (3D CAD, photomontage, animation, virtual reality and so forth) is becoming widespread in architectural design practice. These advances have particularly affected communication and decision-making between clients and designers. At present, the development, design and use of visual simulation material in practice is predominantly based on intuition, prior experience and/or trial-and-error approaches. This area has been largely neglected by research, which has resulted in a lack of an established conceptual framework to guide research, aimed at improving practice in this field. As a consequence, there are diverging views over simulation properties and style of representation that should be used in a simulation. The 'photorealistic' approach advocates increasing simulation properties to improve communication, understanding and credibility of visual simulation material. However, the 'abstraction' approach argues that since human information processing is limited, increasing the amount of detail in a presentation could not only be wasteful, but could also deter participants from the intended message of the communication. In conformance with this view, the emerging 'non-photorealistic' approach promoted the production of abstract visuals, similar in style to sketches, paintings and drawings. However, there is a lack of evidence to support the claims made by these two approaches. In practice, representations varying in simulation properties and style of representation have been indiscriminately applied with a limited concern about the implications for the effectiveness of communication, participation and decision-making. The purpose of this study was to determine the effects of simulation properties and the style of representation on design decision-making of specific architectural design tasks. It sought particularly to establish whether the participants' confidence in the simulation techniques to portray the finished space and building were significantly affected by simulation properties and the nature of the task. As a result, a conceptual model was developed to generate and test hypotheses and guide future research in this field. This model provided the structure and the focus for examining the relationship between the principal components (design-decision tasks, computer visual simulation and decision-makers) and their subset of variables (e.g. simulation properties, style of representation, level of expertise and design training). It sought to investigate participants' responses to simulation properties, their confidence in the simulation and preference for a style of representation. The confidence in their decisions and preferred styles of representation were considered as measures of the quality of the design decisions. The experimental study was designed to provide participants' responses to a controlled variation of simulation properties. The effects of three levels of computer-generated architectural representations on design decision-making were evaluated in a series of experiments designed to measure participants' responses along cognitive, affective and motivational dimensions. Moreover, the decision qualities were assessed through the confidence in simulation and preference for the appropriate style of representation. The confidence ratings were examined according to simulation properties and design-decision tasks. The preference ratings were assessed in relation to the appropriate style of representation and the nature of the tasks. The findings of this study made it possible to infer that while increasing simulation properties was not significant in the evaluation of cognitive tasks (e.g. layout, size of rooms, location, access and scale), it was highly significant in the appraisal of affective tasks (e.g. appearance of building exteriors and interiors). In addition, adding more detail did not significantly affect respondents' expression for the need for change for either cognitive or affective tasks. While, it was found that adding more detail did not increase participants' confidence in simulations during the evaluation of cognitive tasks, the increased simulation properties resulted in greater confidence ratings in carrying out affective tasks. In their selection of the style of representation, participants' consistently preferred the photorealistic representation, independent of the decision-tasks at hand. Nevertheless, these preference ratings were stronger for the affective tasks and were in most cases not related to variations in simulation properties. In summary, the conceptual model proved to a very useful aid of thought and provided a theoretical base to test hypotheses. The results of the research are used to make recommendations for both research and practice for the production and selection of simulation material for design and decision-making purpose.