PEPA has recently been extended with functional rates [1][2]. These functions allow the specification of indirect interaction between components in such a way that the rate of an activity may be made dependent on the local state currently exhibited by one or more components. In this paper we demonstrate that these rates allow a systematic simplification of models in which there is appropriate indirect interaction between components. We investigate the interplay between this style of simplification and aggregation based on bisimulation, and establish a heuristic for applying both techniques in a complementary fashion.