Declarative learning with delayed feedback