Learning from others provides the foundation for culture and the advancement of knowledge. Learning a new visuospatial skill from others represents a specific challenge-overcoming differences in perspective so that we understand what someone is doing and why they are doing it. The "what" of visuospatial learning is thought to be easiest from a shared 0° first-person perspective and most difficult from a 180° third-person perspective. However, the visual disparity at 180° promotes face-to-face interaction, which may enhance learning by scaffolding social perspective taking, the "why" of visuospatial learning. We tested these potentially conflicting hypotheses in child and young adult learners. Thirty-six children (4-6 years) and 57 young adults (18-27 years) observed a live model open a puzzle box from a first-person (0°) or third-person (90° or 180°) perspective. The puzzle box had multiple solutions, only one of which was modelled, which allowed for the assessment of imitation and goal emulation. Participants had three attempts to open the puzzle box from the model's perspective. While first-person (0°) observation increased imitation relative to a 180° third-person perspective, the 180° observers opened the puzzle box most readily (i.e., fastest). Although both age groups were excellent imitators and able to take the model's perspective, adults were more faithful imitators, and children were more likely to innovate a new solution. A shared visual perspective increased imitation, but a shared mental perspective promoted goal achievement and the social transmission of innovation. "Perfection of means and confusion of goals-in my opinion-seem to characterize our age" Einstein (1973) pg 337, Ideas and Opinions.