The article reports a study in which three groups of dyads—mothers and their six-year-old children—interacted in a problem solving task. The dyads had to solve a task that was simultaneously intellectual and manual in the sense that they were to tie a knot (a clove hitch) using a picture as a resource. The mothers differed in terms of their professional and educational background; in one group the mothers were industrial workers, in the second group they were nurses and in the third group they were teachers. The interactions were videorecorded and analysed with respect to who was responsible for performing the necessary steps in the attempts to tie the knot and what responsibilities the child was invited to take on. The results indicate considerable similarities between groups on many variables (such as time used on the task, number of words uttered, etc.), but they also show that the teacher mothers were more inclined to involve the child as a performer and to organize the cooperation in such a way that the child had to engage in the semiotic activity of relating the picture to the tying of the rope. It is argued that the differences observed reflect varying conceptions of what it means to learn in a situation of this kind and what the child should do in order to learn. However, it is obvious that these differences between the dyads do not represent anything that we can conceive of as stable traits or competences. Rather, they seem to a large extent to be produced through differing interpretations on the parts of the dyads of how to interpret the very situation in which they were acting.