Paul E. Dickson, Tim Richards, Brett A. Becker

In the computing education community, discussion is growing about the benefits of teaching programming by explicitly using notional machines to help students. To-date most work is largely theoretical and little work addresses actually using them in a classroom. This paper documents our experience of creating a notional machine for a specific course and using it in that classroom. A key point we learned while creating this notional machine is that many of the difficulties encountered were due to the concept of a notional machine being tightly coupled to students' mental models. Although not surprising, the numerous complications this brings are important to overcome. The potential amount of detail included in the notional machine is enormously influenced by the students' mental models, which are likely specific to a course, and also change throughout a semester -- and certainly across several semesters. We present lessons learned from this experience, among them that implementing a notional machine and using it in class is a non-trivial yet possibly beneficial exercise.