A part of the systematically designed researches on the above-mentioned topic which the writer had carried out for many years was reported. The findings of the first couple of studies on the evelopment of -transposition learning suggest that there are two levels of generalization ; the first level is erceptual level (stimulus generalization) based on stimulus equivalence and the second is cognitive eneralization (mediated generalization) based on mediation process which may be either verbal or non-verbal. To test the development of mediated generalization in preschool chilaren three- and five-year-olds were assigned five tasks including primary and transfer larning of size discrimination, graded in difficulty in terms of smaller difference ratio and increased number of irrelevant cues (1967). They consisted of 3 groups (16 in each) ; G1 was trained to choose "absolutely" the same size in either pair I : 3 or 3 : 5 randomly presented (the numeral referring to the step of size), G2 to choose "relatively" the smaller and G3 to choose also "relatively" the larger in each pair. The absolute choice was found more difficult than the relative choice in both age groups. Those who failed in the primary training were trained with the same stimulus pairs in terms of the classical transposition technique and those who succeeded in this went back to the primary training. The total failure in the primary training was only one case in the "absolute" choice group of the older, whereas the younger groups had more cases of failure. The above results were compared to those with white rats (GI and G3) which were trained in the same way with the same size stimuli. There was no spontaneous sucess in primary training in the animals and no shaped success in the "absolute" choice group, whereas some in the "relative" choice group. As for transfer learning there were failures in four tasks among those spontaneous or shaped successes in the primary training. The results of the verbalization test indicate that those who succeeded in mediated generalization cannot necessarily verbalize the reasons of their choices especially among the younger children.