Previous research has shown that in recognition tasks, a distinctive feature can increase hit rates and decrease false alarm rates associated with an isolated item in a similarity space. However, this is inconsistent with the prediction of the global activation models, such as the Generalized Context Model. Since it is generally assumed that recognition and categorization operate under the same similarity-based generalization mechanism, a distinctive feature should also affect categorization judgments in a similar manner. However, the effects of feature distinctiveness on categorization has yet to be explored. For this reason, the present paper investigates the effects of feature distinctiveness on recognition and categorization, alongside with the effects of isolation and encoding strength. The results of the experiment suggest that the feature-based distinctiveness effect arises in categorization tasks in a way that is consistent with the mirror effect in recognition tasks. These findings bolster the line of literature that categorization and recognition operate under the same generalization mechanism.