Human memory systems are commonly divided into different types of store, the most basic distinction being between short-term memory (STM) and long-term memory (LTM). Phonological STM, as proposed in the working memory model is closely linked to semantic LTM. Nevertheless, the mechanisms of maintenance with STM, and transfer of information with LTM are poorly understood. Candidate mechanisms within phonological STM are rehearsal (either articulatory or elaborative), and refreshing. There is also evidence of long-term learning within STM. In this paper we use the Behavior and Reasoning Descriptive Language (BRDL) to model human memory contents as well as the perceptions that allow humans to input information into STM. By using the Maude rewrite system to provide semantics to BRDL and dynamics to BRDL models, we can explore various cognitive theories about phonological STM maintenance and transfer of information for long-term retention, such as articulatory rehearsal, elaborative rehearsal, and refreshing. This approach has been implemented in a tool that allows cognitive scientists to carry out in silico the simulation of learning processes as well as the replication of experiments conducted with human beings in order to contrast alternative cognitive theories.