Describing the structure of concepts through different feature levels

Abstract

It has been suggested that concepts can be recognised based on at least some amount of highly shared features across individuals (i.e. core features). However, the boundaries between those features and less-shared ones remain unclear. Therefore, the main purpose of this paper is to operationalise the different levels of the structure of a concept's meaning and to test the effect of those levels on semantic decision latencies. First, we performed a feature based cued-recall task. Then, we designed two speeded verification tasks in which features were presented before and after the concept (i.e. feature-concept and concept-feature condition, respectively). Results of the cued-recall task allowed us to operationalise core features. Furthermore, we observed that core features were verified faster than partially shared, followed by features, and partially shared features faster than idiosyncratic ones, being features, and that this effect is independent of the order of stimulus presentation.