Toward a Computational Model of Creativity: Novel Hypothesis Generation from Structural Knowledge

Abstract

Creativity, generation of a new idea from past experience and knowledge, is one of fundamental aspects of inferential process making progress in many scientific and non-scientific fields. Children's learning at their early development needs to be creative: by nature, they frequently encounter new situations in which they need to infer about things unfamiliar to them. In the present study, we attempt to review empirical and theoretical studies on creative inference in children's word learning. Two theoretical implications for creative cognition are discussed. A computational model of word learning offers a formal way to analyze the relationship between hypothesis generation and structural prior knowledge, which can potentially explain some aspects of empirical findings on new idea generation.