According Balacheff and Clement models, knowledge (K) and conceptions (C) interact mutually. Clement seems to define conceptions as a set of items fueled by values, practices and knowledge while Balacheff defines knowledge as a set of situated conceptions. Moreover, Chevallard and Charlot make the hypotesis that learning requires a factor commonly underestimated: the “relation with learning”. This study surveys the impact of the relation with learning on the evolution of scientific conceptions and knowledge among learners. It intends to test the Charlot and Chevallard hypothesis as well as the Balacheff and Clement models in order to identify a possible interaction between knowledge and conceptions. We surveyed a population of gymnasium pupils (n = 47) about blood circulation. One questionnaire enabled to determine a set of relation with learning profiles. Besides, two questionnaires were distributed one month before the learning sequence (phase 1), then two months after the learning sequence (phase 2) : one questionnaire for identifying the conceptions, the other one for measuring the knowledge level. Our results show clearly that the two factors are correlated: utilitarian profile (U) and pleasure profile (P) both improve in the short run the evolution of the scientific conceptions and knowledge. The knowledge evolution and the conceptions evolution are rather similar during the sequence we have surveyed. These data do not disconfirm the interactionnal hypothesis between C and K inside the set circumscribed by Balacheff and Clement. On the other hand our data do not allow so far to assure whether conceptions constitute a set of knowledge or not. These data are relevant to shed light on the learning process and emphasize the neccessity to integrate the “relation with learning” into Giordan’s allosteric model claiming to transform the initial conceptions of the learners.