Motor sequence learning and the effect of context on transfer from part-to-whole and from whole-to-part.

The present study attempts to characterize the contextual conditions (i.e., addition versus omission of elements) that enable or prevent transfer of an acquired skill. The effect of learning and transfer from part-to-whole and from whole-to-part was studied with the serial reaction time (SRT) task. In this study, two alternative sequences of the SRT task were utilized, a short (i.e., 'part') sequence consisting of six elements (ADBACD), and a long (i.e., 'whole') one consisting of 12 elements (BDCADBACDABC) in which the short sequence was embedded. Three groups participated in the study: one was trained with the 'whole' sequence and two with the 'part' sequence (differing in the number of initial training trials performed), for six blocks followed by a random block. Then, for an additional block, each group was divided into two subgroups, one which continued to practice the same sequence, while the other was transferred to the alternate sequence (i.e., 'part-to-whole' and 'whole-to-part'). Results indicated that the group that first practiced the 'whole' and then the 'part' sequence showed full transfer, while the other group showed only partial transfer from the 'part' to 'whole' sequence. The findings of the present study are inconsistent with Thorndike's principle of identical elements, and, instead, indicate that full transfer is enabled in spite of certain contextual changes (i.e., omissions), but only partial transfer is enabled when other changes are applied (i.e., additions).