SWITCH: A Simulation of Representational Change in the Mutilated Checkerboard Problem.

摘要

One of the best ways to study insight is to study problems that require shifts in representation. One such problem is The Mutilated Checkerboard (MC) problem. This report presents a brief psychological account of problem solving in the MC domain, followed by a detailed computer simulation of how change of representation might actually occur. The computer simulation, SWITCH, was build in the production system language, Soar. Analysis of how SWITCH works leads to psychological claims about problem solving strategies, representation, and the mechanism underlying representational change. Insight is that flash of illumination during which a problem solver exclaims 'Aha ' and sees (or thinks s/he see) the answer. To have an insight, one must first have a problem that constitutes something of a puzzle. If the problem is difficult for merely technical reasons (e.g. the following of a long and tedious algorithm), there is little room for insight. However, if the problem requires that one or more of its elements be thought of in a new way, then insight seems quite likely. There is good evidence that insight often follows rapidly after the problem solver represents the problem in a new way (Kaplan Simon in press). Unfortunately, it's quite difficult to collect psychological data at the exact moment of insight, since human subjects are almost universally silent just before the AHA. (KR)