Recent studies in category learning and usage have shown that there are shifts in category representation. In the present study, we develop three models categorization that consisted of different learning objectives to examine cognitive mechanism underlying the representational shifts. The results of simulation indicated that the representational shift observed in Johansen \& Palmeri (2002) can be explained by selective attention, selective exemplar memorization, or mere random chance. Although these models could not be differentiated based on classification generalization patterns, a detail examination of acquired model coefficients were discussed in order to design future studies.