Although a great deal of effort has been put into the research and development of artificial intelligence reasoning systems, spatial reasoning is a relatively new independent research area. Up to now spatial reasoning problems have been considered in a variety of areas, including computer graphics, computer vision, robotics, geographical information systems, man-machine interaction, autonomous systems, and expert systems. Spatial reasoning involves spatial task planning, navigation planning for robots, representing and indexing large spatial databases, the integration of symbolic reasoning with geometrical constraints, and multisensor data fusion.In this paper I focus on the aspects of spatial reasoning that are more closely related to high-level computer vision. More precisely, after a brief review of studies performed by psychologists of perception related to the field, I investigate the problems of: i) the description of objects and space modelling, ii) the representation of spatial relationships, iii) functional aspects of objects and naive reasoning.