Scientific simulation of light distribution in architectural spaces has been used in the past to generate numeric values of lighting performance indicators such as illuminance levels and daylight factors (e.g., on task surfaces), luminance levels, and glare indices. Designers and consultants typically compare such numeric results with minimum or maximum requirements in relevant illuminating engineering standards to decide if a particular design meets mandated performance criteria. Recently, it has been suggested that such traditional numeric evaluation methods may be complemented (or even substituted) by approaches that rely on scientific visualization tools, enabling the users to virtually observe the illuminated space and directly evaluate its lighting. Such scientific visualizations combine photo-realistic rendering with detailed photometric computation, thus providing an image of the architectural space that is a dependable representation of its lighting. Consequently, designers could use computational visualization of lighting conditions in spaces in order to judge their visual quality. We consider this idea as a hypothesis in need of empirical testing. In this paper, we describe such a test.