Four experiments tested a theory of memory and cognition which assumes that verbal and nonverbal information are processed in functionally distinct LTM systems. Subjects presented with pairs of pictures or printed names of animals and objects differing in rated real-life size were instructed to choose the conceptually larger member of each pair. the one that appeared to be farther away. or the one whose name was easier to pronounce. The following results were consistent with theoretical predictions: (1) RT to choose the larger member increased as the memory size difference decreased, for comparisons between as well as within conceptual categories; (2) memory size comparisons were faster with pictures than with words, whereas the reverse occurred for pronounceability comparisons; (3) with pictures, but not with words, size comparisons were significantly slower when real-life (memory) size relations conflicted with physical size relations than when the two were congruent; and (4) the size congruency effect was reversed for relative distance judgments of pictured pairs. These results cannot be easily explained by current verbal coding or abstract (propositional) theories of LTM representations.Moyer (1973) reported a study in which subjects compared the sizes of named animals from memory. The animal names had been previously ranked by the subjects according to their judgments of the relative sizes of the animals themselves. In the experimental task, subjects were visually presented the names of two animals, such as frog-wolf, and were required to throw a switch under the name of the larger animal. The result was that the reaction time (RT) for the choice increased systematically as the difference in animal size became smaller. Specifically. RT was largely an inverse linear function of the logarithm of the estimated difference in animal size. Since this function is similar to the one obtained when subjects make direct perceptual comparisons of patterns differing in size, Moyer argued that subjects compare animal names by making an "internal psychophysical judgment" after first converting the names to analog representations that preserve animal size. Smaller size differences between animals presumably are represented as smaller differences between the internal analogs, and the resulting decreased discriminability is ret1ected in increased reaction times.Moyer's study is unusually relevant to the question of how information about the perceptual world is represented in long-term memory. Although the experiment was not designed explicitly to test such a theory. the results are generally consistent with a dual coding approach according to which verbal and nonverbal information are represented and processed in functionally independent, though interconnected,