本研究主要的目的是在比較基模抽象模式與特殊範本模式在新動作遷移及動作保留上的預測效果。本研究以14名大學一、二年級未學習過任何相闊的體操動作的男生處受試對象，所有的受試者於練習階段必須學習地板睡足躍起、吊環屈仲上九週'每週三次，而後進行新動作遷移測驗(雙槓、單槓屈仲土)及動作保留測驗(吊環、雙槓屈仲上)。本研究所得到的資料以相依樣本t考驗處理後，結果顯示:新動作遷移會受到新動作與先前練習動作間的相似性高低所影響;動作保留上則發現，在經過一段未練習的間隔後，先前練習過的動作還有保留動作參數可供利用，因此會表現的比較好。所以本研究的結果支持特殊範本模式對運動技能記憶表徵的預測。同時也發現，支持特殊範本模式與否，可利用實驗情境來操控，當練習動作的數量少且彼此間的界限分明時，就傾向於支持特殊範本模式;當數量多且界限不分明時，則傾向於支持基模抽象模式。

The purpose of this study was to contrast predictions from the schema abstraction model with those from the specific exemplar model of motor skill memory representation. Fourteen male undergraduate students participated in this study. All subjects were naive regarding the study's purpose. In both experiments， subjects performed neck kip and rings kip of gymnastics over 9 weeks， three times per week of acquisition. Then subjects received transfer(parallel bars & horizontal bar kip) and retention (rings & parallel bars kip) tests. The data were analyzed by t test with repeated measures. The results indicated that novel task transfer was effected by the degree of similarity between the acquisition and transfer exemplars， and there was advantage for a previously produced exemplar over a novel exemplar after I-week retention interval because an action plan containing the specific parameter values would be available. Theseresults were indicative of a specific exemplar model of memory representation for motor skills. Also， this study found that evidence either for or against a specific exemplar model can often be obtained by manipulating of the experimental conditions. Nomally， practice on only a fewer acquisition exemplars that have more distinctiveness results in support of a specific exemplar model whereas a large number of acquisition exemplars that lack in distinctive boundaries results in support for a schema abstraction model.