Causal analysis and inductive learning

PUPS learning mechanisms center around the representation of causal relationships between objects or events. These learning mechanisms involve (1) a set of principles for inducing causal relationships in absence of a domain theory; (2) a set of analogical mechanisms for extrapolating the causal analysis of one situation in order to predict or problem solve in another situation; and (3) compilation procedures for turning these extrapolations into general production rules. These learning mechanisms are illustrated with respect to a detailed example from the algebra tutor where they do a good job of reproducing the instructionless learning we observe of students.