Three empirical studies were undertaken to test the theory of "psychological differentiation" proposed by Witkin, et al (1979). Ninety-four science gifted, 120 music/art gifted, and 85 average students served as subjects. The main findings were as follows: (1) No significant negative correlation between score of left hemisphere specification (LHS) and that of creativity was found, after Torrance´s ipsative scale, Your Style of Learning and Thinking (YSLT), was revised into a Likert type scale. This was contradictory to previous study which said that the score of right hemisphere specification (RHS) is positively correlated with that of creativity, but the score of LHS is negatively correlated with that of creativity. It was pointed out that the result obtained in previous study might be due to the nature of ipsative scale perse. (2) Multiple comparisons after ANOVA and ANCOVA showed that both science gifted and music/art gifted students were more prone to use R-style of learning and thinking, and were more field-independent, than were the average students. This was interprerted as an evidence supporting Witkin´s differentiation hypothesis. The result of discriminant analysis also showed that it was effective to discriminate the gifted subjects from the average ones when scores of YSLT and Hidden Figure Test were used as dependent variables. (3) Results of Hotelling T^2 test revealed that while R-style subjects got higher score on Hidden Figure Test, L-style subjects did not get higher score on Verbal Reasoning Test, as it was predicted by the theory of lateral specification of hemisphere.