The use of categorical data in computer-based land analysis is a signi‡cant challenge because it usually leads to a binary treatment of the information in a subsequent analysis. Cognitive science suggests that humans need categorical data to process experiences, form memories, analyze, or summarize and communicate knowledge (Lakoff, 1987; Rosch, 1978). Similar reasons underlie the common practice of measuring and storing land-cover information as categorical data. We ‡nd it intuitive to talk about “forest cover,” “grassland,” and “sand dunes,” but despite the inherently experiential and subjective nature of these terms, we are able to effectively communicate ideas using them.