Unleashing Creativity: A Machine Learning Approach to Investigating Domain Specificity and Domain Generality

The uncertainty of whether creativity is domain-general or domain-specific influences the design of curricula and training for effectively fostering and developing creativity. To further test this question, this study adopted three innovative approaches: a) focusing on differences in students’ creativity at the between-major level (e.g., music, math), rather than the commonly studied within-person level; b) adding the social science domain to the comparison between art and science, which is lacking in the existing literature; c) using machine learning techniques in addition to classical analysis methods to deal with the data. Overall, no significant differences in creativity were found in students who studied in different domains. Clustering analyses drew possible divisions in students’ creativity across majors but also displayed overlaps between majors. By using supervised decision tree model techniques, this study used creativity scores as well as demographic variables of students to predict their majors. Regardless of how predictors were selected, the model could not accurately predict (accuracy > .50) majors. Overall, almost all results supported the notion that creativity is domain-general. Based on that, this study suggests that educators can pay more attention to designing general programs and curricula to cultivate common/general creativity across domains.