Extending ACT-R’s Memory Capabilities

To resolve several problems of ACT-R’s declarative memory (DM), Schultheis, Barkowsky, and Bertel (2006) developed a new long-term memory (LTM) component, called LTM . In this paper we present two ACT-R interfaces which integrate LTM into ACT-R. Such integrating LTM makes it easily accessible to ACT-R modelers and allows more thoroughly evaluating it in its interplay with other components of a cognitive architecture. By considering four different memory phenomena we show that ACT-R with LTM is superior to ACT-R employing only DM and, thus, (a) LTM ’s benefits are not impaired when integrating it into a cognitive architecture and (b) using the newly developed interfaces improves ACT-R. In particular, integrating LTM into ACT-R allows computationally exploring memory conceptions which cannot be modeled with ACT-R utilizing only DM.