The paper explores the plausibility to favor a productive use of cognitive resources activated in one context to solve a problem in another one. The study was based on a manifold view of cognition considered as the result of the context-sensitive activation of cognitive resources, as proposed by Redish (2004) and Hammer (2004). Nine university physics students were involved in the study. They were interviewed in groups while solving two problems dealing with geometrical optics. Results show that including a particular problem helps students activate resources that are productive to solve another one in which those resources are not spontaneously activated