This paper gives a method to determine a person's hypothetical conceptualization of an object -- its breakdown into subassemblies, subsubassemblies, and so on -- from the person's sequence of requests for pieces used in constructing it. A technique is given to determine whether, given a group of conceptualizations, there is a typical one. The hypothesis that assembly instructions presenting a typical conceptualization will yield better structural and functional performance than those presenting a minority one is supported experimentally. Conceptualizations are derived from objects built from memory (and incorrectly) by people who first studied typical or minority instructions. A new distance measure determines how far these conceptualizations are from those presented in the instructions. People studying typical instructions yield typical conceptualizations, and importantly, people studying minority instructions also yield typical conceptualizations, although they are significantly less typical than those from typical instructions. From the theoretical construct of conceptualizations and the methods of measuring them a practical principle, and a way to implement it, are found. The principle: When a single set of procedural instructions is designed, it should present the conceptualization that the majority of people to be instructed by it bring to the situation naturally. (Author)