The majority of the models which attempt to explain category-specific deficits are based on the assumption that the conceptual knowledge is represented in a permanent way in memory (abstractive view). Carbonnel, Charnallet, David, and Pellat (1997) showed that a non-abstractive view would be more suitable to account for some of these cases. The present study aims to assess the hypotheses on the meaning evocation process as formulated by these authors, using a property verification task conducted on healthy subjects. The data were in agreement with these hypotheses and would be more hardly accounted in the frame of the abstractive view.