PRIMs is a cognitive architecture based on the popular ACT-R cognitive architecture. The goal of cognitive architectures is to simulate and predict human cognition, and to provide a basis for intelligent systems.
The most important aspect of PRIMs is that it is aimed at creating models that go beyond a single tasks. This means looking at transfer of knowledge between tasks, how task knowledge can be constructed on the basis of existing skills, and how our structure of skills is bootstrapped in development.
The central idea in PRIMs is that control of information processing can be carried out by primitive operations. These primitive operations control the flow of information between different cognitive modules, such as vision, long-term memory, task goals, working memory, etc.
Each of these cognitive models communicates through a set of buffers. Together these buffers are conceptually similar to the idea of a global workspace.