Abstract. Models created using artificial intelligence (AI) methods can provide novel perspectives on theories of moral development and yield new insights into human moral judgment. Using latent semantic analysis (LSA) and the machine learning method k-nearest neighbors (k-NN), we investigate moral judgment as characterized by the Standard Issue Moral Judgment Interview and Scoring System (Colby & Kohlberg, 1987). The resulting model consists of a collection of textual representations for moral schemas generated using LSA and k-NN from prototypical criterion judgment responses in the scoring system. Preliminary results demonstrate promise for the approach and suggest at least one way the computational approach could augment a previously identified challenge to human scoring.

Full poster (pdf)