- This study examined mirror and non-mirror arm selection processes in an observational learning context. Observer groups watched either a novice (instruction or discovery) or skilled model performing a bimanual task with the right arm leading the left arm. The models were viewed from a third-person perspective. Observers of the skilled model more often selected a mirror-image (left-hand) hand-lead in post-observations tests, while observers of the novice models more often selected a non-mirror image (right hand) hand-lead in post-observation tests. This is a novel finding regarding arm selection processes in a learning context, yet it is consistent with imaging data that has revealed specific neural areas linked to the selection of mirror and non-mirror imitation processes for first- and third-person viewing perspectives. The skilled model also supported more accurate and stable performance of the bimanual task in observers compared to the instruction and novice models. It is concluded that a skilled model supports attention focus being directed at pattern analysis, while novice models support attention focus being allocated to strategy identification first, followed by pattern analysis.