Schema acquisition from a single example.

Three experiments were conducted to compare similarity-based learning and explanation-based learning in schema acquisition.Similarity-based learning approaches hypothesize that concepts are formed based on similarity among multiple examples.Explanation-based learning approaches hypothesize that a general schema can be acquired from a single example by connecting instantiations of existing schemata and generalizing the connected portion of the example.The present experiments demonstrated that subjects could acquire a plan schema from a single example in knowledge-rich domains as predicted by the explanation-based approach.These experiments also showed that subjects were not able to carry out explanation-based learning if they did not have sufficient domain knowledge and if the schema to be acquired was not structured by causal constraints.