Creativity is often considered a static trait, but recent work has shown that a creative state can be induced through explicitinstruction to be creative. A two-term verbal analogy generation task (e.g., GLOVE : HAND :: : ) that includeda randomized instruction to Answer Creatively vs. Answer Quickly was used to explore the impact of state creativity,and convergent and divergent thinking upon the creativity (i.e., semantic distance) of the generated analogies. Resultsconfirm that instruction to Answer Creatively yielded more semantically distant analogies. Additionally, the magnitudeof improvement between instructional conditions was predicted by performance in the Quickly condition. Participantsproducing less creative analogies in the Quickly condition benefited substantially, whereas participants producing morecreative analogies benefited less. Convergent and divergent thinking predicted more creative analogies in the Quicklycondition but not in the Creatively condition.

Analogical reasoning entails both one-to-one alignment and relational transfer. Yet the relative reliance on one processover the other may depend in part on the extent to which role-based relational reasoning is available. We systematicallymanipulated two theoretically important item characteristics that impact the extent of role-based relational reasoning insolving semantically distant verbal analogies: (1) the analogical relation (composition vs. category coordinate), and (2)the format of the analogy stem (i.e., two vs. three terms). For the categorical analogies (WATERMELON : PINEAPPLE:: VELVET : SILK), stem format had no effect. Whereas for the composition analogies (WATERMELON : SALAD ::VELVET : DRESS), participants were faster to solve the 3-term than the 2-term analogies, thereby indicating a facilitativeeffect of role-based alignment (e.g., both watermelon and velvet as materials of their respective objects). Thus, resultssupport analogical models positing the detection and use of relational roles (Holyoak, 2012).