Elementary schoolchildren of second and fourth grade were given two types of successive concept identification task, that is, pictorially and verbally presented tasks. In each 64-trial task, feedback was given on every fourth trial in order to probe S's hypothesis used (Blank-trial method). Interaction between grade and task was found for using hypothesis, organizing hypothesis by the stimulus dimensions, and selecting the correct hypothesis quickly. Second graders were less efficient in the verbal task than in the pictorial one, while forth graders showed equally efficient performance.