We describe an improved version of a cognitive model that exhibits the expert to novice transition in solving physics problems. The initial model was written by Larkin and initially translated into Soar (version 4) by Levy. In revising it to run in the latest version of Soar (7.0.0), we have updated it to be an exemplar of an understandable and reusable cognitive model: it includes graphic displays for understanding how it works, and it has had its learning mechanism organised as a general learning utility for use in other models where further principles can be specified. It is available via anonymous FTP. We argue that this example standard of displays and reusability must be more often realised if cognitive modelling is to prosper.
[Appeared as : F. Ritter & G. Baxter (1996) Able, III : Learning in a more visibly principled way, Proceedings of the 1st European Cognitive Modelling Workshop, Berlin]