ABSTRACT
BETWEEN the first stage characterized by graphic collections and the third stage, that of hierarchical classification based on logical operations, we find a second stage in which the characteristic response is the non-graphic collection. We use the term “collection” rather than “class” in the strict sense, because the former term carries no implication of a hierarchical structure of class-inclusions. However, these collections are no longer graphic, and objects are assigned to one collection or another on the basis of similarity alone. Nevertheless, these several collections are simply juxtaposed, instead of being used as the basis of a hierarchical class structure. As we saw in ch. I, non-graphic collections are already foreshadowed in the successive associations of stage I, since very often they engender a series of similar objects, i.e. a non-graphic collection in posse. But it is unusual at this stage to find the similarity being respected when the subject turns to a consideration of the collection as such. During the course of stage II, non-graphic collections become more and more the rule, and it is important to see why. Anticipating the results and their discussion, we shall simply note that their growing predominance is largely due to an increasing differentiation of intension and extension, and hence a greater co-ordination of the two. We can hardly attempt a complete analysis of the process involved without first considering the quantitative relations to which it gives rise, and this forms the chief topic of ch. III, where we shall be studying the question of how children eventually come to differentiate “all” and “some”. For the present, we will have to confine our account to a broad general description of the way in which classifications are evolved.
