Previous research (Goldstone, 1994) has proposed a model (Similarity as Interactive Activation and Mapping, or SLAM) of similarity assessments for structured scenes. In this model, the parts of one compared thing must be aligned, or placed in correspondence with the parts of the other thing. The primary assumption of the model is that the calculation of similarity involves an interactive activation process whereby correspondences between the parts of compared scenes mutually and concurrently influence each other. Three predictions of the model are empirically tested. First, unlike most other models of similarity, SIAM can predict a nonmonotonic relation between featural overlap and scene similarity. Second, SIAM predicts that the presence of shared nondiagnostic features increases accuracy in forming alignments between scene parts. Third, with reasonable assumptions relating judgment difficulty to information complexity, SIAM predicts that the number of objects in a scene has a large impact, relative to the number of features in an obiect. on subiects' difficulty in judging scene similarity.