From Causal Models to Analogical Inference

Keith Holyoak
University of California, Los Angeles (UCLA)

Analogical reasoning enables the generation of inferences based on relations shared by as few as two cases. A challenge for the Bayesian approach to induction is to formalize how analogical inferences are generated and assigned probabilities of truth. Problems to be resolved include (1) the representation of relations, (2) the computation of approximate isomorphism, and (3) the connection between causal models and analogical inference. Some preliminary experimental evidence bearing on these issues will be presented.

Presentation (PDF File)

Back to Long Programs