Part and whole practice: chunking and online control in the acquisition of a serial motor task.

A four-component aiming movement was used to examine the relative effectiveness of part and whole practice. Following a pretest, participants were assigned to one of three practice groups. Participants in a "Whole" group practiced the four components together as a unit. A "No Overlap" group practiced the first two and last two components of the task, alternating every fifth trial. An "Overlap" group practiced the transition between the second and third components on every trial by alternating practice of the first three and last three components every five trials. Participants in all groups improved significantly from pretest to immediate posttest and maintained their performance over a 24-hr delay. Contrary to the "chunking hypothesis," participants in the No Overlap group improved as much as those in the other two groups. Kinematic data indicated that participants in all three groups learned to use response-produced feedback earlier in the individual movement trajectories. Moreover, participants appeared to acquire a general ability to make transitions between movement components rather than specific transitions. The results suggest that segmented or segmented "overlap" practice regimes may benefit learning movement sequences of short duration.