Obstacles to the skill-based approach: Why is skill reuse so difficult for cognitive architectures?

Skill reuse is a commonly accepted aspect of human cognition but it has been difficult to translate to cognitive architectures. We developed the skill-based approach which enables modelers to create models composed of skills created for other tasks but it does not (yet) support fully reusable skills. We will discuss three factors that prevent full reusability: inflexible WM, rigid goal selection and all-or-nothing condition checking. The factors are discussed in the context of the architecture PRIMs but they also apply to many other cognitive architectures. Finally, we discuss possible solutions to alleviate these issues.