Views and Causality in Discovery: Modelling Human Induction
Our recent efforts have involved modelling “instructionless learning”. This is a sort of induction that human learners exhibit when they are asked to “figure out” complex systems (e.g., programmable toys) without instructions or advice. Instructionless learners often undertake major reorganizations of their theories but may not, in the process, lose the syntax and semantics of particular parts of their theories. How do these reorganizations take place? They do not think very carefully about changes in their theories and therefore rely heavily upon the analysis of errors. How does error recovery work?
Unable to display preview. Download preview PDF.