Much research in digital spatial technologies has had the implicit goal of eliminating humans from the analysis process. This has not been successful because humans are able to carry out many tasks which are difficult to program (e.g. pattern recognition). It is appropriate to re-examine the role humans should play in spatial information systems. Technology development should be oriented towards combining human and computer processing in such a way that each assists the other by carrying out the tasks that each does best. Ways of handling spatial data based on these ideas are discussed and a new spatial data representation based on uncertainty is presented. Design issues related to integrating human analyses into computer processing are discussed and a remote sensing technique which embodies these ideas is presented. It is concluded that existing spatial data handling needs to be rethought and restructured in order to ensure continual human participation in the analysis and decision-making process.