Representing Goals Modally: A Production System Model of Problem Solving in the Tower of London

Abstract

GLAM-PS (Glamorgan Problem Solver) is a production system model of problem solving in the Tower of London (TOL) puzzle. It is introduced as a draft cognitive architecture that is similar to John Anderson’s (1998, 2004) ACT-R, but represents system goals, long term memory and production memory modally, rather than amodally. The current paper demonstrates how GLAM-PS models problem solving on 3-disk TOL problems (a comparison with human data is also made). GLAM-PS uses representations of intended actions to control behaviour and planning utilizes the simulation of future problem states


Back to Friday Posters