A consensus has been reached concerning the fact that in the observation of learning processes, the individuality of learning is to be considered, but the practical application of individualising learning in classroom teaching still raises a big problem. For although in this field, many aspects concerning the individual learner (e.g. previous knowledge), the learning process and the subject matters are being examined, but often the connecting of results (from these examinations) is omitted. Moreover, in most studies, one finds statements about the “average learner”, but one can hardly draw any conclusions about the individual learner from them, i.e. the individual learner is not the focus of interest in these studies. This raises the question for the individual’s learning. There are several other concepts of the terms of learning paths or learning ways (“understanding processes” cf. Niederer (1997), “learning paths” cf. Gelmann, Rochel; Brenneman, Kimberley (2004), “learning ways as a succession of subject matters”, learner’s methodical approaching of the subject matter cf. Hempel (2002)). Thus, it is interesting to explore the possibility of a more extensive construction of learning ways, thereby considering the following unsettled problems:  How can individual ways of learning be described by means of learning decisions and depending on the learner’s personal attributes as well as the learning success?  Is it possible to identify typical individual learning profiles?  What is the teacher’s function in an open multimedia learning scenario?

In order to be able to answer these questions, we developed some open learning environ-ments, implemented with the aid of a multimedia CBT-programme and an experimentation box. Our learning environments are characterised by a range of options to choose from as wide as possible. According to Deci & Ryan (2000) these freedoms of choice provide the learner with a high motivation to deal with subject matters. In our learning environments the choice of options applies to several levels (context, medial access (e.g. film, experiment, learning tempo). In order to be able to maintain the complexity of the research conditions, it seems suitable to choose the concept of “Design-based-Research” (cf. DBR 2003, Gorard/Taylor 2004) One aim is the description of as many conditions as possible. For this, qualitative as well as quantitative means are used. Among these are, besides tests concerning previous knowledge/learning success, cognitive abilities, self-concept, especially screen-video and video recordings, interviews and learning diaries, which aim at recording the learners’ decisions.

So far it turns out that there are significant individual differences in the use of learning environments, e.g. some learners try to get a general idea about subject matters first of all, while others stick to a subject at once and work on this intensely. As to the choice of media, it turns out that real experiments as well as their film versions are almost always used by nearly every learner. As far as the self-concept is concerned, it turns out that up to the present, learners with a weak self-concept achieve better learning success than those with a very strong self-concept, but the data analysis is not finished yet. We also conclude that our kind of learning environments cannot work without the presence of a teacher.