Redirecting learners' attention during training: effects on cognitive load, transfer test performance and training efficiency