Previous research has shown, that when 10 year old children trained to become experts in solving a sequential problem are placed in a situation of transmission to an untrained receiver (novice) of the same age, they interact through verbal messages and gestures using tutoring procedures adequate for informing the novice and regulating his procedure when solving the problem. We have analyzed (coded) the communication on the functional level using a list of behavioral categories that are combined to form "communicative events". The behavior records were then analyzed structurally with the Theme program. In this research, the cognitive capacities of the receiver were not controlled and the expert had to discover and adapt to these during the interaction. We have used the same method but before entering the dyadic situation the receivers (novices) were presented to the experts as efficient vs. inefficient thus creating in the expert social representations regarding the receiver. Each expert had to deal successively (in a counter-balanced order) with receivers that were characterized as "good communicators" or "not very good". This indication corresponded or not (i.e., for half of the subjects) to the results of a pre-test carried out among the future receivers. In this way four expert-novice groups were defined. The functional and structural analyses carried out in an identical fashion show significant differences between the experimental groups. In particular, in the cases when the novices do not correspond to the level indicated, one sees, by comparing the beginning and the end of the interaction, that the level discovered during the interaction is taken into account. We will present in a contrasting fashion the behavioral patterns found in each of the groups at the two indicated moments. This illustrates the role of the Theme program as a "Structuroscope" allowing the detection of fine structural modifications in dyadic interactions.