Background: Generalization is a vital aspect of real-life motor learning. We asked whether in a realistic skill (bean bag throwing) generalization occurs within or beyond the range of trained movements and whether this is different for constant or variable practice. Methods: what was your outcomes? How you measured them? In two experiments participants threw beanbags at a target at various distances. In the first experiment (n=24), two training groups threw beanbags to a constant near or far target and were examined at an intermediate transfer test. In the second experiment (n=80), participants trained either at a single target (constant), or two targets alternatingly (variable) with targets placed at different distances and they were tested for transfer within and beyond the training range. A control group was included which only performed the transfer tasks. Results: For the near transfer target, no group outperformed controls (P>.05), whereas all groups except the near constant group (P=.072) performed better than the control group at the intermediate target, and only the far constant training group performed better than controls at the far target (P<.02). Conclusion: These results suggest that generalization is limited in this task. The generalization that was found depends mostly on the distance between the training and the transfer target, not on whether the transfer target is within the trained movement range. The superiority of the far constant group over other groups further suggests that the farther away the goal was, the greater the need for specialized training.
