While commercially available mixed-reality (MR) head-mounted devices are also increasingly used in education the impact of MR learning environments is mostly being evaluated with respect to learning outcomes and learning gains alongside a number of affective variables. Here we aim at a deeper understanding of the influence of MR on experimental problem-solving processes by means of eye-tracking analyses as an addition to self-reported and questionnaire-based test instruments. On the basis of a didactically and empirically sound designed MR learning environment for experimental problem solving with an electrical circuit we investigate how novices and experts differ in their success in problem solving and their visual attention for both a MR learning environment and a non-MR condition. Our results indicate that MR not only has a positive effect for both groups on success in problem solving, but that it is also accompanied by significant shifts of visual attention on the different parts of the experiment.