Key points are not available for this paper at this time.
SUMMARY We compared five distinct classes of models of how people judge the relative sizes of named objects. The model that posited simultaneous processing of both analogue and discrete representation (as denned herein) was most consistent with the findings of our four experiments. In the first experiment, people learned the sizes of six stick men, each of which was a different size and was drawn in a different color. Further, people categorized the three larger figures as and the three smaller ones as small. After learning sizes, colors, and categories, people received pairs of color names and decided as quickly as possible which referred to the larger figure. The amount of overlearning of category labels critically determined whether the disparity in sizes of compared figures always affected decision times; only when category labels were very overlearned was the size disparity effect (wherein figures more disparate in size are compared faster) partially eliminated. In the second experiment, the same basic procedure was used to study the effect, wherein two relatively large things are compared faster in terms of which is larger than in terms of which is smaller, but vice versa for two relatively small things. The congruity effect was partially eliminateb1 when category labels were highly overlearned. The results of the first two experiments were inconsistent with predictions from models that assume only one kind (analogue or discrete) of representatio n or that assume that different types always are processed sequentially in an invariant order. The last two experiments investigated the role of visual imagery as a form of analogue representation. These experiments provided evidence that images are used when to-be-compared objects fall into the same size category; when objects are in different categories, the category labels themselves can be (but are not always) used to perform size comparisons.
Kosslyn et al. (Thu,) studied this question.