Performance modeling has been made easier by architectures which package psychological theory for reuse at different levels. Both CPM-GMS, which packages theory at the task level, and ACT-R, which packages theory at the lower level of rules for perceptual-motor interaction, have been shown to be useful. This paper describes ACT-Stitch, a framework for translating CPM-GOMS templates and interleaving theory into ACT-R. The research involved in producing ACT-Stitch will benefit reusable template research by showing how to implement templates and interleaving in a new architecture that processes resource information. ACT-R research will benefit from re-usable productions packaged at a higher task level and from the multi-tasking control structure used that allows ACT-R to interleave productions from different templates. The zero-parameter predictions of ACT-Stitch are empirically validated.
