From implicit skills to explicit knowledge: a bottom-up model of skill learning

This paper presents a skill learning model CLARION. Different from existing models of mostly high-level skill learning that use a top-down approach (t hat is, turning declarative knowledge into procedural knowledge through pra ctice), we adopt a bottom-up approach toward low-level skill learning, wher e procedural knowledge develops first and declarative knowledge develops la ter. Our model is formed by integrating connectionist, reinforcement, and s ymbolic learning methods to perform on-line reactive learning. It adopts a two-level dual-representation framework (Sun, 1995), with a combination of localist and distributed representation. We compare the model with human da ta in a minefield navigation task, demonstrating some match between the mod el and human data in several respects. (C) 2001 Cognitive Science Society, Inc. All rights reserved.