Recently there has been a revival of interest in visibility analysis of architectural configurations. The new analyses rely heavily on computing power and statistical analysis, two factors which, according to the postpositivist school of geography, should immediately cause us to be wary. The danger, they would suggest, is in the application of a reductionist formal mathematical description in order to ‘explain’ multilayered sociospatial phenomena. The author presents an attempt to rationalise how we can use visibility analysis to explore architecture in this multilayered context by considering the dynamics that lead to the visual experience. In particular, it is recommended that we assess the visual process of inhabitation, rather than assess the visibility in vacuo. In order to investigate the possibilities and limitations of the methodology, an urban environment is analysed by means of an agent-based model of visual actors within the configuration. The results obtained from the model are compared with actual pedestrian movement and other analytic measurements of the area: the agents correlate well both with human movement patterns and with configurational relationship as analysed by space-syntax methods. The application of both methods in combination improves on the correlation with observed movement of either, which in turn implies that an understanding of both the process of inhabitation and the principles of configuration may play a crucial role in determining the social usage of space.