Working memory and acquisition of implicit knowledge by imagery training, without actual task performance

Abstract

This study investigated acquisition of a mirror-reading skill via imagery training, without the actual performance of a mirror-reading task. In experiment I, healthy volunteers simulated writing on an imaginary, transparent screen placed at eye level, which could be read by an experimenter facing the subject. Performance of this irrelevant motor task required the subject to imagine the letters inverted, as if seen in a mirror from their own point of view (imagery training). A second group performed the same imagery training interspersed with a complex, secondary spelling and counting task. A third, control, group simply wrote the words as they would normally appear from their own point of view. After training with 300 words, all subjects were tested in a mirror-reading task using 60 non-words, constructed according to acceptable letter combinations of the Portuguese language. Compared with control subjects, those exposed to imagery training, including those who switched between imagery and the complex task, exhibited shorter reading times in the mirror-reading task. Experiment II employed a 2 x 3 design, including two training conditions (imagery and actual mirror-reading) and three competing task conditions (a spelling and counting switching task, a visual working memory concurrent task, and no concurrent task). Training sessions were interspersed with mirror-reading testing sessions for non-words, allowing evaluation of the mirror-reading acquisition process during training. The subjects exposed to imagery training acquired the mirror-reading skill as quickly as those exposed to the actual mirror-reading task. Further, performance of concurrent tasks together with actual mirror-reading training severely disrupted mirror-reading skill acquisition; this interference effect was not seen in subjects exposed to imagery training and performance of the switching and the concurrent tasks. These results unequivocally show that acquisition of implicit skills by top-down imagery training is at least as efficient as bottom-up acquisition.