Task-specification language, or theory of human memory?

The authors' theory plays a dual role as task specification language and abstract symbol level theory of human memory. As a specification language, it seems useful as far as it goes, but it cannot specify the training half of the task. As a theory of human memory, it posits data structures and processes while abstracting away from control structure. But the set of primitives is incomplete (no learning), and may posit an implausible relationship between recognition and recall.