In the paper we discuss the relation between fuzzy sets and the graded membership and typicality effects found in the study of concepts. After a short overview of the topic, we present three experiments, carried out using the same method but with different situational contexts, which examine whether graded membership and typicality could be considered as independent factors capable of influencing the performance of human participants involved in sentence verification tasks, or they are somehow interrelated. The paper concludes with a general discussion of the experimental findings and the problems they pose for models of concepts based on the theory fuzzy sets.