Skill Transfer Observation based on Rule Embedding for Skill Understanding: An Artificial Grammar Example

摘要

For the purpose of human skill understanding, we use artificial grammar task to create a situation in which we know the underlying rules supporting given skill and can observe a skillful subject's behaviour. To achieve artificial grammar embedding, the subject is asked to classify strings as grammatical or ungrammatical. The experiment continues with observation of a transfer attempt in which the subject explains the newly acquired skill to another, naive subject. The results reveal that the subject discovered useful string decompositions without having any prior knowledge about the grammar. These decompositions rely on intuitive feeling rather than on explicit rules.