In 1991, an international workshop held in Bremen was working on " Research in Physics Learning-Theoretical Issues and Empirical Studies " (Duit, Goldberg, Niedderer 1992). The intention of this workshop was to develop a new research goal, which was to study learning processes with data from during the learning process. The main point, which often makes learning process studies very important and interesting is the following: Students' actual constructions are often different from taught knowledge. In other words: there is a gap between what we teach and what is learnt (McDermott 1991), the knowledge to be taught is different from students' steps of learning (Tiberghien 1997). The students undertake a cognitive development, which leads them towards constructing certain "intermediate conceptions" (see below) corresponding to their cognitive structure (Niedderer 2001). In this paper, we are presenting some theoretical and methodological issues together with some results from new studies (Givry 2003; Budde 2004). 2. Theoretical framework Conceptions and expressed ideas Many authors use the term "conception" to denote their basic concept of thinking and learning (see Duit 2004). A conception is seen as a hypothetical set of statements, skills, procedures, that the researcher attributes to one or more students in order to account for students' behaviour in a set of given situations (Tiberghien 1997). Here, we distinguish between two cases: (1) One is to consider that a conception intends to be part of modelling students' mind. It is a construction of a researcher to describe typical use of elements of knowledge and ways of thinking of students. A conception has to be stable and must appear in more than one context and point in time. (2) The other is to consider that the researcher infers only ideas, which are expressed in the students' productions without making hypotheses on students' mind (Givry and Roth in press). These inferences will be called " expressed-ideas " to distinguish them from the previous approach. These two approaches show that, even if the points of view on the students' mind modelling are different, the analysis and the results on learning are compatible and mutually reinforced. Both approaches are inferred from students' productions (utterances, gestures, writing) by the researcher and can show some stability over time (Niedderer 2001) and through several situations (Givry 2003). The set of these situations represent their domain of validity, this domain can be reduced to a single situation or be stable in several situations. In both cases there is a construction of a researcher, which describes the core of several ideas of students in the researchers own words using the most distinctive features of those ideas for this description. This procedure is also aiming at a considerable reduction of data, describing the core of a set of ideas in a set of situations with one conception. Learning processes Often, learning processes 1 can be represented as a sequence of conceptions developed by students during instruction. These conceptions do not exclude each other; a student can have 1 The use of the term process here can be justified by stating that it describes a series of steps, which allow to acquire the scientific taught concepts and then are a student's way of going from initial to final conceptions ("learning pathway").