Background: Research indicates that animations presenting procedural instructions lead to better learning if the animation displays the procedural task from a first-person perspective (over-the-shoulder) compared to a third-person perspective (face-to-face). Aims: This study extends view-perspective research by investigating whether the observation of human hands completing manipulative tasks in an animation are necessary or not. Sample: Sixty university students participated in the study. Method: Participants studied two knot-tying animations from a first-person perspective showing hands, or a third-person perspective showing hands, or a first-person perspective without showing hands. Results: Results showed that studying first-person perspective animations resulted in higher performance on a knot-tying task and recognition task (but not transfer task) than studying the third-person perspective animations. The strongest effects were gained from the first-person perspective animations showing hands, although comparable learning outcomes were often found with the no-hands perspective animations. In addition, spatial ability was found to influence knot-tying and recognition performance, while gender minimally interacted with performance in the different viewing perspective conditions. Conclusions: Hand-manipulative task are learned most optimally from animations when presented from a first-person perspective, while it is not necessary to show the hands.